... humming machinery is fine, amplifiers not so much ...

update 19-Nov-2021
- Add Debug Workbook - Bias Mods


The M-125 is incredibly good value in today's insane HiFi Jewelry marketplace. It's not pretty or child safe, but is sonically very, very good. The amplifiers take everything in stride. It's FUN to listen! Reviewed by Glenn Young and enthused users comment favorably in the Dynaco Tube Forum*, AudioAsylum, diyAudio, etc. The amp is very versatile with both pent/triode modes, and is sonically configurable with a variety of tube combinations. Since my Citation II I've had a soft spot for KT‑88 amps and owned several. In pentode mode, the M-125 travels a middle ground between the PrimaLuna Prologue 5 [PL5] and NuPrime ST-10. It's not as recessed as the PL5 nor as forward as the ST-10. IMO, I'd adjudge the amp a perfect match for the ≈4Ω TC-50s when paired with Kimber BiFocal-XL in my system and ROOM!

March 2020 - Spica TC-50 retired for tri-amped Eminent Tech LFT-8b. NuPrime ST-10 for woofers, M-125s mid and PL5 tweeter. 2 x miniDSP 2x4HD for crossover. Kimber Monocle-XL for the mids and HannLinte Braided OFC 10AWG for woofers and tweeters.

As always, YMMV.

Mr. Latino was very responsive to pre-purchase email questions about the amplifiers.

This missive is intended for the competent DIY enthusiast. If you purchased built amps, by all means have at it if you feel competent.
 • ieLogical M_125 Ultralinear assy manual. Version 28-Nov-2021-b to build the VTA / Tubes4Hi M-125 as an Ultralinear only version with improved bias, 4 power tube bias check... and other assorted improvements. Format follows original build instructions, so you may pick and choose. Please contact ASAP with any anomalies.
 • VTA M_125 Debug Workbook to assist in troubleshooting.

From the Tubes4HiFi website [current as of 14-Nov-2021]:
Power output (4 X KT88) 125 130 watts from 17 Hz to 32 kHz at less than 1% THD in ultralinear - 65 70 watts in triode
¿¿¿ Somehow the amps got more powerful with the same transformer specs ???
- Nope. 1kHz 125W distortion is 2.8%. 1kHz distortion reaches 1% @ ≈55W.
Power output (2 X KT88) 70 watts from 17 Hz to 34 kHz at less than 1% THD in ultralinear - 40 watts in triode
- probably not a good idea to run w 2 power tubes. Could be hard on the heaters.
IMD ........................... < .38% (19 kHz and 20 kHz) at rated power
- Nope. CCIF 19/20kHz d2L is 0.82% @ 6W, IMDpwer is 0.84% @ 0.6W
Frequency response ......... 10 Hz to 35 kHz +/- .1 dB at 1 watt
- Nope. -0.19dB @ 50Hz & -0.1dB @ 11kHz relative to 1kHz
Power bandwidth ............ 17 Hz to 37 kHz +/- .3 dB at 125 watts
- Nope. -0.27dB @ 20Hz & -0.87dB @ 27kHz relative to 1kHz
Sensitivity .................... 1 volt in for 125 watts out
- Nope. ≈1.6v which is somewhat beyond most -10db consumer gear.
Feedback ..................... 10 dB
- Nope. ≈3.9dB
Version ..................... ??? - currently v19 as of 19-Sept-2021
- Nope. PCB layout identical to v15. v16 dropped ground plane fill. At best, these are point revisions, i.e. Circuit layout is the same. Some amps delivered with v16 component values and v19 boards. Mid 2021 schematic has a minor change in component values.
Mid 2021 v19 schematic changes relative to early 2019 v16:
  • bandwidth reduced from 96kHz to 31kHz !?!?!? c12 changed to 680pF
  • • a bit more feedback - r8 reduced to 6.8k
  • • a bit less gain in the phase splitter - r20/22 changed to 27k
  • • a bit lower HP filter to phase splitter - c6/8 changed to 0.33µF
  • • it should be note the above two changes alter circuit voltages, but since these are not accurate, par for the course
  • • OP transformer? Above changes could be to compensate for a new transformer, but if so, one should expect sonic deltas. Deltas are not a negative per se, as long as they complement the rest of the system. No component operates in isolation!
In short, who knows how the new amplifier performs or sounds. M-125 gain was already low for unbalanced -10dBV consumer levels. On line specs not updated as of Sept 19, 2021. !! CAVEAT EMPTOR !!

Claims for power consumption, SNR, distortion and feedback are grossly inaccurate and likely never verified. Almost certainly not verified for each of the plethora of arbitrary modifications over the years.
We did. Driver board revision v16. See the Pudding @ lab on the cheap. Load: 8Ω  16nH
Amps drive Kimber Monocle XL 2.5m & LFT8b mid panel: 10.7Ω  37nH...

... which just goes to prove specs are often not worth the paper to print!


Lethal V

WARNING: Lethal Voltages!!

Click thumbnails for a larger image.
ESC or Click Upper Right 'X' to close.


pimple Switch
Two heavy boxes arrived, one with an ominous pimple. The assembled amps were shipped in the kit box, too small for such heavy objects, permitting only woefully inadequate packing. Upon removal, the left amp rattled. The power switch toggle had been forced into the switch, destroying it. The impact force was enough to cross-thread and deform the mounting nuts, making removal particularly difficult.
Sample Sample Wiring
One of the deciding factors to purchase factory assembled M-125s was the quality of the wiring as shown images on Tubes4HiFi and The Dynaco Tube Audio Forum. The wiring delivered is not quite up to the others. Additionally, driver board and tube complement may be somewhat arbitrary. The M-125 web page, which appears to be well out of date, stated “The recommended tube complement is ... 2 X 12AU7 or 12BH7 driver tubes”. Driver board revision v16, [Why did v16 drop GND plane on v15?] w U.S. Sylvania-made NEW National 5963, a data tube similar to 12AU7, received March 2019. Opinions vary as to 5963 suitability as an audio tube, but like everything audio, opinions are only valid in the system under test. From a construction standpoint, US made tubes are insuperabile. In the M-125, they are excellent wrt PL Chinese 12AU7.

In response to a query regarding the 5963 substitution, Mr. Latino replied “In the last 9 years running the M-125s with 5963 driver tubes...” which some might construe Bait and Switch recommending one tube and delivering another.

Additionally, the 12AU7, 12BH7, 5963, 5814, 6189 or CV4003 Plate Characteristics vary considerably. Changing willy-nilly upsets bias, gain structure and distortion.
Interestingly, the 12AU7/5963 requires ½ the heater current of the 12BH7. Tube heater supplies are nominally 6.3vac ± 10%, with ±5% providing longer tube life. At 117v line voltage, with 5963/12AU7 tubes & 2xKT−88 heater voltage is 104% of nominal. v1/3 are 106% of nominal. v2/4 heater supplies appears designed for 12BH7 drivers with 2x KT−120s. If line voltage is anywhere near U.S. standard 120v, a bucking transformer is required with standard [2x 12AU7/5963 & 4x KT-88/6550] tube complement. Schematic, mechanical and parts list to build the pictured Bucking Transformer. Optionally, but not optimally, a variac may be used. Beware of cheap Chinese products recommended by Messrs. Latino & Mottram.
Other tube combinations will increase the heater imbalance, possibly reducing SNR. TWO tube, half-power operation requires further line voltage reduction to keep heater supplies within 10% of 6.3v nominal with standard tube complement.

Mid 2020: Amps now ship with 115/120V primary transformer, obviating the need for a Bucker or VARIAC.

Kit builders estimate about 10 hours to build an amp. There are less than 100 solder connections. An experienced builder making multiple sets with pre-cut lengths should easily complete a pair in about 14 hours. The factory cost to build is $817 or about $50/hr, a not unreasonable rate for an experienced technician, plus about $100 for shipping packaging.




After repairing the broken switch, connecting up the amps, powering them on, I was greeted with a buzzy hum!?!?!? The hum, audible from 1m, did not diminish with only the input shorted amps connected to the TC-50s, not particularly efficient @ 86dbV.

Mr. Latino stated “My impressions are that you are 'over analyzing' the amp and its circuitry when maybe you should just be listening to the music.” Right. On amps with less than 55db SNR!

The system is minimal: CD6006 −> LC1 −> XO1 −> Amp. LC1 & XO1 are self-designed passives. See Last Hurrah. AC power is dedicated circuit w 10ft run to AC panel. I have multiple amps and ALL are dead quiet. The only Earthed component is/are the power amp[s].

As a bit of background, my company, Studio City Sound Corp, designed and installed recording studio wiring in Los Angeles. With monitor level on S T U N, all one could ever hear is a slight thermal rush. Buzz? Humm? _ N E V E R !! _ Ditto my HiFi for the past ½ century. We designed scoring panels for film consoles, The Vocal Splicer, distribution amps and other tidbits. I was head of Analog Design @ AMS-Neve, perhaps the world's premier recording console manufacturer, from 1996 to 1998. HiFi grounding is child's play compared to a 96 input analog console, 2x24 track, multiple 2T recorders, hundreds of microphone inputs, outboard channels, tri-amped main monitors, 3 sets of small monitors, multiple cue systems interfaced across multiple control rooms and studios. I consulted w Monster Cable, Ibanez, Alesis, Yamaha and others wrt the sonic quality of their products. I also designed satellite based system monitors for ocean going freighters, ferries, tugs and large yachts. Bad grounding at sea can be fatal to both personnel and ship.

After several fruitless “It's not the amps” emails, I declined Mr. Latino's offer to return the amps.

surely you're joking, Mr. Latino?

Assembly Faults

Hand-built amps are not always pretty. These, however are just plain UGH-Lee! It is appalling how poorly the amps were assembled by Ben/Benjamin “I take pride in my soldering and quality work is of utmost importance to me” Woloss, Tubes4hifi and VTA Amp Tech.

Amateurish. Slipshod! Unsafe!! Garbage!!!

ShortLeads GndLug Solder Nutz
WireJoin BadCap pinch pinch

I would have been much further ahead had I returned them. However, they do sound very good and seemed to have potential once quieted.

I do like a good challenge:

  • Left Amplifier Faults
    • • AB-Q board linked to driver with short crossed leads making service impossible
    • • AB-Q R20-C6 lead across trace on solder mask, wire not in hole
    • • fuse lead hanging wire, cold solder joint
    • • Earth Safety cold solder joint
    • • dry joint on Green Heater CT
    • • nicked 6.3VAC power wires separated from AB-Q board leaving stripped end in hole when de-soldered
    • • spliced GND lead on supplementary cap module
    • • AuthentiCap loose, not well secured by tabs and prone to accidental chassis short
    • • output transformer v2/v4 500VDC blue lead pinched between transformer and case
    • • dry joint V2/3 fell off when adding snubbers

  • Right Amplifier Faults
    • • power transformer bolts not tightened, barely finger tight, 2 nuts not down to chassis
    • • VERY POOR Earth Safety ground crimp
    • • GND 'star' bolt barely snug - multiple GND lugs should have shorting jumper
    • • fuse Holder not tight, about ⅓ turn loose
    • • ESL cap solder to fwd GND lug, leaving long uninsulated lead exposed which could touch GND on AuthentiCap
    • • AB-Q 6.3vac GND lead nicked and broke when de-soldered to fix polarity
    • • v2 bias lead nicked, broke on AB-Q removal
    • • dry joint on NFB lead which showed up after about 150 hours
    • • dry joint on r14 robbing v6 of current
    • • R22-C8 lead to ABQ board fell off

  • Both Amplifier Faults
    • • AuthentiCap GND cold joint
    • • dull grey solder joints indicative of low-temperature/dirty iron and/or movement during cool
    • • insufficiently flowed solder 'tacks'
    • • inconsistent assembly: some leads wrapped around terminals, some pass through, some just tack soldered
    • • inconsistent wire lengths, twist, connections and solder joints
    • • bias measurement jacks not snugged - required ½ turn to tighten
    • • speaker binding posts insufficiently tight. Rotated when finger tightened. Required ¼ to ½ turn to tighten.
    • • spliced AB-Q leads
    • • inconsistent and wrong heater wiring for use in 2 tube ½ power mode
    • • melted insulation
    • • overly long wire strips leaving uninsulated 500VDC B+ unnecessarily exposed
    • • uncleaned driver board flux blobs
    • • speaker Black nuts stripped, plastic rotates and will not tighten on spade
    • • teflon insulated wire for AB-Q. Teflon is very difficult to strip w/o special stripping tool.
          That several leads broke indicates incompetence/wrong tool
    • • 20k & 30k resistors swapped starving v6 and applying 300vdc to v7 rated 250v MAX

Mr. Latino did not respond to email reporting assembly faults.

Design Faults

    • • Power Tube Schematic REVERSES C6 & C8
    • • Multi-point GND:
      • ⏚ AuthentiCap
      • ⏚ soldertab 1
      • ⏚ soldertab 2
      • ⏚ Earth Safety ground
      • ⏚ driver board mount
      • ⏚ driver board GND loop
    • Nichicon PT series driver capacitor voltage ratings are too low.

      Rated 400VDC. Max applied voltage is 510VDC w/o tubes and 455VDC w tubes on 115VAC line. Higher on 120VAC. Replaced w Rubycon 100µF 500V 10-12k Hrs @ 105°C.
      either a valve rectifier or B+ delay MUST be used. Rubli HTR214 Rev D highly recommended. ≈27s delay and no contacts to degrade, unlike Tubes4HiFi / VTA offering

    • • v7 LM334 CCS lacks temperature compensation.

      Location causes L/R imbalance due to temperature drift. LMM334 case temperature oscillates 45-55°C over several minutes, concomitantly changing phase splitter current flow and thus, gain.

    • • choke wiring poorly routed and fields reinforce
    • • DC wiring is cheap PVC. 20ga 600v 150°C Silicone preferred. 300v OK in a pinch.
                Amazon.com : silicone +20 gauge +solid -stranded
    • • AC BIAS wiring run next to DC
    • • unterminated tube rectifier heater winding
    • • 120VAC power wiring adjacent to DC
    • • long audio input lead run through choke field
    • • long feedback lead run through choke field
    • • feedback and input leads run parallel
    • • mismatched power transformer polarity
    • • ESL capacitor location
    • • v1&3 tubes should be mirrored and HT leads run between pairs for best CMRR
    • • speaker jack binding post mount inadequate for heavy cables, jacks rotate
    • • power tube sockets should use flat head screws or be atop the chassis to preclude KT‑88 rocking
    • • non-standard 16mm speaker terminals spacing - 3 crammed into space for 2 - precluding better quality WBT or similar
    • • Mode Switch contact rating is overkill for required current. Performance may degrade from non-use
    • • feedback taken from 16Ω tap, compromising 8Ω & 4Ω distortion
    • • marginal 75µF SCM
    • • marginal sub-chassis ventilation
    • stability components omitted
    • no output transformer no load protection
    • likely reuse of ST-120 power transformer which has two heater circuits with balanced loads. M-125 has two heater circuits with unbalanced loads compromising 60Hz noise

how bad can it be?

LeftStock LeftStock
Per Mr. Latino, pre-purchase [February 2019]:
“I had someone test the M-125s about 9 years ago. He came up with a -95 dB signal to noise ratio. I do not know how that was 'weighted'. Typically with good input driver tubes, you have to put your ears on the grille cloth to hear any hum/noise with speakers of average efficiency.”

All measurements made on 4Ω tap. Unweighted SNR ≅ 55db L / 51db R. Measurements made with shorting plug, amps connected to 4Ω 100W load resistors, both amps plugged into same grounded outlet.
Either amp alone was somewhat quieter. Not quite ear on the grille cloth quiet, but better.
With amps connected to source, hum/noise almost tolerable at seating position 4m distant.


LineSyncAdpt BuildStand
Line Sync Adapter: Gives a 100:1 reduction for use with scopes with reduced input range and/or do not have Line Sync. DO NOT use this before determining the power socket polarity is correct. A mis-wired socket or power strip would put line voltage on the ground connection. If the voltmeter does not measure ≈1.2vac S T O P !! and sort out the fault.
This jig should only be used for initial transformer and heater polarity determination. Once polarities are determined, use one of the heater voltages referenced to Audio GND.

Build Stand: A pair of stands make working much easier. Heights are ex-pads. [From similar on The Dynaco Tube Audio Forum]

Power Transformer Polarity

TransChoke TransPolarity SyncLeft SyncRight
First thing to check is the power transformer polarity. Since the Audio GND is tied to Earth Safety, if the transformers are inverted, induced fields are inverted and may increase noise. I've been checking polarities since the '70's and it often pays dividends to correct.
Polarity can be detected by means of a choke placed next to the power transformer and the choke output measured with a scope synced to the line frequency. Choke lead polarity is arbitrary, but marking it keeps measurements consistent. Measure only one amp powered at a time. As can be seen, the power transformers were inverted relative to one another. Choose the less noisy polarity and mark the power transformer leads, fixing as necessary.
Subsequent to my amp delivery, VTA entered the mid-20th century supplying amps with 115 and 120VAC power transformer taps. That should ensure correct polarities, but it never hurts to check. I recently used a small 250CT transformer, Hammond no less, with secondaries wired in phase. In other words, 0VAC between the taps. Neither I nor Hammond had ever seen similar and all Hammond transformers are 100% tested. Hammond replaced it gratis.
The drawing at left shows induced current flow with transformers with inverted and correct polarity.
The lower block shows the effect of a correct Single Point Audio GND. As built, with the 9 point Audio GND there is resistance between each gain block. Ground currents shift the reference. With a true single point ground, the relative voltage at each gain stage does not shift.

Heater Wiring

The build instructions correctly advise inverting heater polarities between v1 & v3 and v2 & v4 sockets. No mention is made of heater polarity for the driver and AB-Q installation.
A feature mentioned on the website: “NOTE - The amp may also be operated with just TWO OUTPUT TUBES - one of v1/v3 output tube and one of v2/v4 output tube.”
Given that there is no provision for determining output heater polarities, arbitrary polarities on any tube pair could compromise SNR. If half power operation is planned, ensure heater polarities are the same for socket pairs v1/2 and v3/4.
The image shows the Heater Polarity concept for a Push-Pull [PP] amplifier. The inverter stage splits the signal into inverted polarities. If Heater Polarity is inverted, it functions the same as the signal and creates a current in the Output Transformer. If Heater Polarity is the same, no signal is created in the Output Transformer.
NOTE: For pairs of tubes, e.g. 4 output tubes, heater polarity should be inverted between the tubes. Tubes in mechanically the same location, e.g. v1/v2 on the M-125 should have the same heater polarity.
For most tubes, improvement is slight as their construction accounts for heater fields. One caveat in two tube ½ power operation: Tubes will see higher heater voltage. Raising the A/C heater voltage with some tubes by 0.5V doubles the hum injection. Another caveat is raising the heater voltage shortens life. In an all out design, use an external regulated DC heater supply.

Heater Polarity Chart

P is in-phase with line, N is out.
Polarity is best determined before the transformer heater wiring is connected.
The right amp v2/v4 heaters were in-phase when they should be out, contributing to the as built 4db SNR reduction.
It was determined by inspection that Modified polarities produce the most benign noise signature in the M-125s I received. Other Revisions might require other choices.

I do wish Tubes4Hi / VTA were less circumspect in publishing revision changes.
With 16 M-125 revisions in 10 years, it's almost Amp of the Month Club!

DC Wiring

DCwiringStock WiringOrig
Prototype at left looked reasonable. Wiring as built very poorly implemented. AC input tie-wrapped to GND and DC. Choke input twisted with output contaminating the output with the pulsed DC input.
Chokes are not counter-placed to cancel fields. Chokes sit directly BELOW the Output transformer. Minimizing field radiation pays off.
Too much back and forth to the capacitors unnecessarily increases lead radiation and induction lengths.

DCwiringMod AuthCapMarked
Mount the chokes with leads inboard. Wire from the rectifier-out/AuthentiCap terminal to the rear choke leads. Wire the choke output direct to the AuthentiCap. Link the AuthentiCap to the Supplementary Cap Module [SCM]. Connect the output transformer RED lead to the SCM. Wire the linked AuthentiCap & terminals and the added black lead to the SCM input. Wire the ESL film cap across the input and output of the SCM.
Space may be tight with the Pent/Tri switch installed.

Miscellaneous Wiring

RectHeater ACWiring AudioInput
Rectifier Heater: With a Ted Webber WS1-T solid state rectifier, rectifier heater wiring is unused. An unterminated winding is a potential noise source. If using a solid state rectifier, remove 5VAC rectifier heater leads from socket, add a ≈5k/1W to terminate the winding and mount off to the side to remove 120Hz pulsed 500VDC into the winding.

AC Input: With externally controlled systems, a power switch is superfluous. It's never a good idea to run 120VAC leads in close proximity to DC or signal wiring. Hard wire power transformer to IEC socket and route away from all other wiring. If a power switch is required, mount it on the rear panel. This will cut hum substantially.

Audio Input: Without the power switch, there a convenient hole on the front panel for the Audio Input jack. In addition to short leads, the audio input no longer passes through the choke fields and maximizes distance from the AC power fields.

GroundLifts/CheaterPlugs are DANGEROUS, Irresponsible and in some locales, ILLEGAL!!

Single Point Ground

Although the M-125s are MonoBlocks, functionally they become one amp when connected to a common source due to their unbalanced inputs. With the Earth Safety ground connected to Audio GND, Ground Loops are almost a certainty.
Rod Elliott has a very good page on Earthing and Ground Loops in particular. The bottom line is that there should be only one Audio GND point. FULL STOP! The Earthed chassis should function as a Faraday cage only and ne'er the twain shall meet.
As designed, the M-125 has a 9 point ground. Never good.

CapInsulation CapCutout AuthentiCapInsulatorBottom AuthentiCapInsulator
AuthentiCap: Other than cachet, why use it? Small value caps have much higher ESR. Sticking multiple caps in a can reduces the temperature rating / life at temperature. The impedance of different value parallel capacitors is exceedingly complex and may contribute to observed instability. D U M B !!

Add clearance for GND tabs. Insulate it from the chassis. Mount it securely with a clamp! Bend the ground lugs inward and add a shorting link across two, connected to the Audio GND Star terminal strip. Wire transformer Audio GND to this link.

GND Star Point: Replace the ground lugs with an insulated terminal strip similar to the one used for the triode mode resistors. Add a buss through the lugs. Link to the AuthentiCap GND and solder all ground leads here.

BIAS Wiring: Remove BIAS lead from wire loom and run down middle of power tubes. The pulsed ½ wave rectified DC has no business loomed with Audio DC.

Driver Board Mount: Replace the 4/40 steel hardware with #6 [3.5mm] nylon screws and stand offs. If using the AB-Q board, its mounting holes need drilling to fit #6/3.5mm hardware.
McMaster-Carr #6 for isolated 0vA:
  •  #6-32 x 3/4 Hex SO 92319A658  •  #6-32 x 3/8 Pan Head 94735A729
  •  #6-32 x 5/8 Pan Head 94735A732  •  #6-32 x 1/4 Hex SO 92319A610

Driver Board GND Loop: Verified M-125 Driver v16 revision ONLY. The driver board ground runs around the outside of the board, the very definition of a loop. From inspection, it appears the loop to the right of the vertical trace is for bias and the left for audio. Cut the loop to reduce the effect of the shared bias/Audio GND. Be certain and careful IF doing this revision!!
Other versions likely have a similar fault, but the cut point may be different.

making it safe

Per the Loop Breaker schematic on Rod Elliott's page, add a 50A [10x device 5A fuse] diode bridge linking Audio GND to chassis/EarthSafety ground. Cap provides RF sink. 10Ω/5W resistor is somewhat arbitrary, with values from 10 to 100 ohms generally regarded as satisfactory.
Amazon: 2x 50A diode bridge Parts Express: 10R resistor  •  100nF capacitor.

Space is tight, so it is necessary to enlarge the right-rear output transformer chassis hole slightly to 11/64' [4.5mm]. A #8-1' bolt is required. Link Audio GND at the heater ground buss. Tie the Earth Safety link at the same bolt as the AC input Earth Safety.

things are looking up

LeftMod LeftMod
Unweighted SNR ≅78db L/R. Still not quite the 95db someone claimed to have measured. The waveform however is a dead giveaway for the half-wave bias. Shame on you, Messrs. Mottram & Latino. A full-wave bias would reduce current pulses by 50% and double the frequency decreasing the noise by 12db w bias LPF mod below.

An additional benefit would be obtained having a separate bias center-tapped winding rather than half-wave tied to the HiV CT [Audio GND].

be careful, your bias is showing


Note that some schematics have incorrect C17 & C18 RefDes.

R39, shown here as a link, should be 1kΩ to improve noise performance. Better filtering can be achieved by changing R39 to 470Ω and inserting 470Ω between D1 and C17 as shown below.

The correct AB-Q connection point is the C18 end of R39.

BiasNoise BiasNoise-6dB BiasModR39b BiasCCT
Bias noise without AB-Q can be reduced substantially by adding a 1kΩ resistor between D1 and C17. The shown position is for clarity. The diode and the resistor should be flat against the board.

BiasV4 BiasV4_Post
Left image is v4 bias as delivered. A classic ½-wave rectification with minimal filtering.
Holger @ Erhard Audio, U.S.A. agent of the AB-Q AutoBias, was kind enough to tell me that replacing R39 on the driver w 100Ω would allow enough current for the AB-Q board and a little more LP filtering. Move the bias lead to the C18/R39 junction.
Optionally, a better filter can be implemented with 2x 470Ω, one replacing R39 and the other inserted between the D1 anode and C17. This should probably also work for standard bias without AB-Q. See Power schematic.
Right image LP filtered bias is 13db better - but not as good as a full-wave regulated DC supply.
Maybe later... arrived. ALMOST. See Regulated Bias below.

AutoBias is EXCELLENT! All tubes hold idle @ -0.500v -0.001v/+0.000 and do not drift. With steady state signal applied, they hold precisely. A much better AutoBias implementation than on the PL5. AutoBias was not measured as built, only on the corrected build.

it doesn't get any better than this

LeftModBias LeftModBias
Unweighted SNR ≅93db L/R. I have to put an ear ON the grille cloth at the woofer to hear anything. Additionally, notice how little hash is present. Absolutely NOTHING can be heard from the tweeter, a vast improvement!

Mr. Latino did not respond to a friendly offer to tell him how the amps had been quietened.

ok, maybe a little better

AB-Qwiring BiasModInstall BiasModComp BiasCct
The AB-Q board handles Bias. There is ZERO need for any Bias on the driver board. Construction of a well filtered bias supply mounted off the driver board reduces bias noise.
Power tube current is returned to the AB-Q board. AB-Q 0vA should be returned to 0vA star. First image is recommended AB-Q connection, wires and parts to remove. Second image is bias supply to AB-Q board.
AB-Q current requirement is very low, ≈11mA. 3rd image compares simulation Bias noise of factory, first 100Ω mod and the filter in the 4th image. A 26db improvement over factory! -55VAC can supply at least 55mA.

regulated bias

RegBiasCct RegBiasAssy
My Nan cautioned 'Idle hands are the Devil's playground.'
So, with my volunteer day jobs of SEAcoast Technical Director and DAV Volunteer Driver up the CoViD-19 spout, I needed some occupational therapy. Hence, fully regulated bias supply. Send me a S.A.S.E. and I'll send you a couple of 1N5622 68V Zener Diodes.

SNR is now >100db unweighted wrt full power.

ieLogical Revision

ModLeft Completed ModRight BiasSelector
ModSch BPmountOrig BindingPost M-125_BP_Mount
Since the amps are now Pentode only, a DP ON-OFF-ON switch was added in place of the Triode resistor terminal strip to allow v1/2 or v3/4 bias verification. Power Schematic - w AuthentiCap symbols, SCM and ESL values and AB-Q connections - as modified from original.

I got fed up with the Binding Posts and mount that would not hold. Replaced with Vampire BP-HEX from Parts Connexion. Mount is milled nylon, but could possibly be 3D printed. These amps remind me of a Fiat 131 mirafiori I owned in the 70's. Once I got the crappy original parts off, it was a great car!

'Audio Zealots'† will understand that the background is now inky black. Delicate sounds shimmer, decay goes longer, impact is more stunning.

Very Nice Amps, Bob & Roy. Truly!

... Original page ended here ...


Jan 4, 2020: created a DTAF topic
    'teaching the VTA M-125 to sing'

3 lines:
  • my VTA built M-125 amps were noisy
  • it is possible to make them almost 40db quieter
  • a link to this page

4 images showing the noise as delivered and as quietened

Jan 5, 2020: Not wholly expected, banned from DTAF. Topics & posts deleted. I guess Smilin' Big Bob wasn't amused.

* - One has to wonder about DTAF comment validity when accounts are banned for anything critical and unrelated topics deleted. Subsequent to publication, many Tubes4HiFi product owners have written confirming a lack of transparency and support from Messrs. Latino and Mottram.

† - Mr. Latino referred to me as an Audio Zealot because I didn't like a PoS Chinese Variac he recommended. In reply, I detailed numerous faults by whom I consider competent, not fan-boy 'it lights up' Amazon users. Posts deleted.

who ARE those guys?

Pre-purchase Mr. Latino stated:
“The sonics of the amp will be the same with the AutoBias board. You won't really hear any difference in sound quality with the AutoBias board installed. I personally have never compared the two (manual bias and AutoBias) MYSELF but others tell me that the sound is the same. The AutoBias board will give you greater convenience.”

All AutoBias products and references scrubbed from Tubes4Hifi.

Mr. Latino states on diyAudio in VTA M-125 mono block kit. on diyAudio:
AutoBias Tubes4HiFi
“Ian's amps have the 'autobias boards' that are sold by an independent distributor. Ian requested that they be added to his wired amp. The -95 dB signal to noise measurement for the M-125 was done 10 years before the autobias boards were ever created. When you remove the old bias pots and 'piggyback' a second board on top the driver board and start adding wires here and there to connect the autobias board, I am sure that the amp's residual noise level will increase somewhat.”
This is utterly false. The product is branded Tubes4HiFi. I found the AB-Q boards on Tubes4HiFi and its propaganda website DTAF. Tubes4HiFi / VTA were building my amps. The original wiring is a rats nest of wires here and there specified by someone who never gave routing a thought! See DC Wiring above.

In response to a pre-purchase query about the AutoBias boards, Mr. Latino stated:
“The cost for just me buying the two autobias boards for the two M-125s are $180. To install them would be $100 for the two boards ($50 per each amp) - Total > An extra $280 for the cost of the two boards + installation. The cost is high. It takes time after installation of the board to adjust the boards for the proper bias”

Yeah, right. All of 30 seconds per amp to adjust level in a shop with a 6.3vac supply.

In response to to a query about R39 specific to AB-Q, Mr. Latino stated:
“I can tell you that on the M-125s R39 can be either a 1K resistor or a jumper wire. I use a 1K resistor but a jumper wire will work OK also. What R39 does is that it sets the 'range' over which the bias may be adjusted.”

Mr. Latino exhibits complete ignorance on something as simple as a voltage divider and LPF relative to the AB-Q install. VTA's instructions are at fault as proven by the 15db noise reduction when the AB-Q is properly installed. See be careful, your bias is showing above.

Why the complete turnabout from the 6 pages extolling AB-Q products on DTAF AutoBias board on Roy's website?

I'm beginning to think Tubes4HiFi / VTA may be good at copying 1950s designs when 117VAC did not have a Earth Safety connection, but incapable of designs for US Standard Earth Safety Grounded [since 1959] 120VAC [since 1967]. Their designs may work in some, but not, all systems. They may be completely out of their depth in solving issues from bad instructions and design.     “Easy? Help — Hard? Ignore”

Mr. Latino stated in the above referenced diyAudio thread:
“I am now not recommending the addition of the autobias boards to any of the VTA amp kits or wired amps.”
Based on experience to date, a Tubes4HiFi / VTA non-recommendation may be a ringing endorsement.

Adding Insult to Injury: DTAF users are not informed that an account is banned. User messages to banned users cannot be replied on DTAF. Users may think the banned user is giving them the shine.
Messrs. Latino & Mottram: Truth will out!
<Soto voce: Should these guys decide to go into politics, their integrity is well on its way.>

See Erhard Audio AutoBias Modules for all the reasons why AutoBias is preferred. These are my 9th tube amplifiers, 2 x M-125 counting as one. The last two are AutoBias and, perhaps due to my advancing age, I'd not buy an amplifier without it!
The AutoBias boards now have updates to make them more bulletproof to get around some of the Tubes4HiFi flaws.

ieLogical has no affiliation with Erhard Audio or AudioAmp.eu.
All information here is presented as a public service to victims of Tubes4HiFi / VTA mistakes and/or non-support.

the project that wouldn't die

ResLeft ResRight NoFB R22-C8
Things started getting weird in February 2020. I was primarily working on Tri-Amping the Eminent Technology LFT-8b, never listening closely, but something was clearly not quite right. Channel imbalance, strange distortion.

Finally getting the amps on the bench, I find the 20k & 30k resistors are swapped on the driver board! A mere 30% over voltage and current on v7. Think Big Bob will send me new tubes? No? Didn't think so either. The NFB lead on the right amp is a dry joint and failed. Instead of Negative it became NO Feedback. Right Amp: R14 also had a dry joint dropping v6 power. R22-C8 lead popped off upon ABQ removal. Nothing to do but tear this garbage to bits and rework it all like I should have done last year!
Upon inspection, I find the soldering similar to crap Chinese, thus every joint was resoldered. Problems solved.

Without a doubt Mr. Bennihaha Woloss is the absolute worst technician klutz I've come across in 50+ years in HiFi.

Ever since day one, the left amp would occasionally fail to play. Swapping cables and tubes appeared to fix, but problem always re-appeared. Hopefully the last factory defect, the left input jack had an intermittent short!!!

RED plated! now I'm pissed

   Shitty engineering drives me batshit!
   For a few pennies, the design could be 100% stable.
   Tubes4HiFi / VTA are either too cheap or incompetent.
   In either case, they should up their game!

After correcting the driver board, the bloody amps red plate and oscillate at anything over ≈40W! Ken Galer had similar problems with his custom build. See M-125 Build on DTAF. The amps are works of art.

Since the M-125 is basically a rip of the Dynaco 120 / Mk VI, I went back to the originals to see what Latino & co. had screwed up. See Dynaco Transformers for the original 120 and Dynaco Mark VI Manual.

Both the 120 & Mk VI have:
  • • power tube plate & g2 snubber resistors
  • • local feedback from the -phase U/L g2 tap nested inside the global feedback loop
  • • a load across the output transformer 16Ω tap [680ohm;/120 - 620Ω/MkVI]. Mr. Latino admonishes never turn the amplifiers on without a load.

GalerNum1Unstable Snubbers Snubbers SnubberInductorIE


Every tube electrode has an associated LCR property. Tubes willingly operate out to 10s of MHz. Parallel tubes can become oscillators when the electrode LCR along with the circuit LCR interact just so across a short length of wire. Power supply may also affect oscillation as seen in diy threads when PSU capacitance is changed. The Tubes4HiFi / VTA M-125 as designed is only marginally stable when built exactly as instructed and parts that hew to the tolerance center. Change wiring, PSU capacitance, layout or tubes and all bets are off. Modification Margin? None!
Dynaco amps with parallel output tubes inserted 47Ω at the electrode in output transformer leads to isolate the electrode capacitance and prevent oscillation. Adding snubbers makes the amps 100% stable. NOTE: My amps are UltraLinear ONLY!! Snubber not examined for Triode mode. USE CAUTION if implementing Triode mode. Triode mode w snubbers schematic.
Adding 20-30 turns of 28ga magnet wire around a 2W 47Ω flameproof 700V Carbon Film resistor essentially removes the resistor at audio frequencies. The 2 to 7µH creates an additional pole that is far outside the open loop bandwidth. 28ga chassis wiring ampacity is 1.4A, far more plenty for ≈75mA/KT-88. Carbon Film has a negative temperature coefficient so the resistance drops slightly with increased temperature, compensating slightly for the increased OPT positive thermal coefficient. Send me a S.A.S.E. and I'll send you 12 feet of 28ga magnet wire.

LM334 Temperature Compensation

LM334RatioGraph LM334_mod LM334_assy LM334_Inst
DriverMod3 M-125_v16_cct_0v
LM334 constant current sources [CCS] are temperature sensitive. Its sweet spot is 20µA to 1mA. The Tubes4Hi / VTA implementation is 10x ideal current which causes a LOT of self heating. It is not temperature compensated and oscillates between ≈45°C and ≈55°C in a 16°C garage. R28 is plugged for close enough @ ≈50°C, a stability the device never achieves. In addition to the not inconsiderable 14°C self-heating, another 20°C is supplied by the tubes after about 30 minutes. With amps side by side, Left amp reaches design value sooner due to heat from Right amp power tubes. In a cool room w airflow, Right amp may never reach design parameters. Space amps adequately to ensure similar values. In a warm room with restricted airflow, the LM334 may come perilously close to 70°C Tmax. ALL Tubes4HiFi / VTA designs using the LM334 for the Phase Inverter tail suffer thusly: long warm up time, temperature sensitivity and balance instability. This can be remedied very easily.

Due to the LM334 instability, the Phase Inverter gain constantly changes, up to ½dB, and as a consequence, the image wanders. The amps never reach equilibrium.
At a minimum, move the LM334 to the underside of the driver board. Note the orientation of the LM334. The flat side of the part faces the flat side of the silkscreen, top or bottom. Mounting under the board will prevent a some of the tube heating, helping to stabilize the LM334. The current may not achieve design values quickly as the internal chassis temperature rises slowly.

For the Temperature Compensated Full Monty:
The diode must be as close as possible to the LM334, ideally in contact for maximum compensation effect. Thermal epoxy the two together for maximum tracking [and minimum serviceability]. NOTE: LM334 V terminal DOES NOT CONNECT TO PCB!!!
Since my amps are built, I opted not to cut & strap the driver board. If starting from a bare board, drill the PCB to more firmly mount the parts. ALL CCS parts go on the PCB underside as shown. The R10a resistor in the image is trimmed to ≈190Ω with a 1k47 on the component side to compensate for 22Ω rather than an ideal 20.5Ω / 240Ω combination. None the less, the amps are now very stable. [These amps are now officially proto-hacks.]

With compensation, the LM334 stabilizes rapidly and does not drift. Sonic image now remains rock solid. Phase splitter balance is surprisingly good, tracking perfectly from naught to clipping.

final thoughts

The deeper I looked into the M-125, the more I questioned many aspects of the M-125 amplifiers. [The entire Tubes4HiFi / VTA series???] From my and others' redplating experience, the amplifiers are only marginally stable. The ONLY redplating tube amplifiers I've ever seen are M-125s and my tube experience began in the 1950's with my Dad's Bogen!! The Tubes4HiFi / VTA amplifiers are ostensibly reworked Dynaco units, yet crucial stability components of the amplifiers were omitted. As mentioned, other that visual cachet, why use the AuthentiCap?

CapBank CapBank2 CapBank3
The capacitor bank is too small. In today's world, Big C comes in little packages, so it seemed wise to increase C. 6 x 330µF 500VDC caps fit nicely in the space of the SCM and T/P mode switch. The bank is built in 3 stages. First make a 1 x 3 bank spacing the caps with 0.0625 thick temporary spacers. Next, build the bank as 2 x 3, again with the spacers. Third, add 0.125 thick insulating standoffs to the bank bottom to allow the caps to vent, should failure occur. Attach with a drop of Krazy Glue, being careful to not run any down the side and glue in the temporary spacers. Spacing is mandatory to allow thermal expansion and cooling.

It is imperative that balancing resistors are installed as in the SCM.
Recommended practice is to use a pair of resistors per cap pair rather than a single pair with capacitors commoned mid-bank.

We tested several capacitor brands. All capacitor brands tested within 1% after forming. We reviewed a half dozen capacitor manufacturers' balance recommendations, some of which needlessly waste power for close tolerance parts.

We used this formula from p11 in Aluminum Electrolytic Capacitor Application Guide from Cornell Dubilier and then quartered the calculated 1.6MΩ value to 470kΩ for additional safety margin. 470kΩ resistors dissipate ≈0.15W, so 1W or larger for safety.

60Hz dropped about 10db in the noise signature with Big C. Capacitor bank secured with 2x bolts [1/4-20 or 6mm] with acorn nuts. A dab of Loctite to ensure permanence.
The 1µF ESL polypropylene capacitor was removed as it did not enhance the sound. IMO, impedance adjustment is more than hanging a film cap across electrolytics. Performance wise, it made ZERO difference.

R_Done BigCmount Bleed R Chassis Spacer
• M-125s take feedback from the 16Ω tap w/o a 16Ω output. The closer the source, the more accurate. Feedback moved to 8Ω tap. Distortion drops about 33% across the audio spectrum.
• Moving feedback to 8Ω, reduces feedback by 0.707 so R8 is changed to 3.9kΩ. This increases feedback slightly to 4.4dB from 3.9dB, 6dB w v6/6829 & v7/12AU7.
• Changing R8 to 3.9kΩ raises the HF roll-off to 185kHz, probably a bit high given the open loop transformer measurements. Changing to 470pF lowers to 85kHz, a bit less than 96kHz stock.
• 220pF C12 placed across R4 to add input 74kHz Low Pass.
• M-125 has no output transformer no load protection. Add 680Ω 5W resistor from 16Ω tap to 0vA for protection in the event of a loudspeaker driver failure, connection loosening, etc. from potentially taking out the OPT.
• A century after tubes became commonplace, controversy over the need to delay B+ until electrodes are hot still remains. Thousands of tube articles and opinions, but never proven. If anyone has proof, send me the link! The Rubli HTR214 Soft Start insertion point is after the AuthentiCap 40µF. The HTR214 has no bleed resistor as the manufacturer expects those messing about inside a tube amp to be safe. 24 hours after depowering, the SCR tab is still at 300VDC. Add 750kΩ as shown to bleed the cap in about 5 minutes.
Always keep one hand in your pocket until you've verified ALL CAPS ARE DISCHARGED!!!
The life you save is your own.
• Before buttoning up, add 6 each 0.25" x 0.125" high round rubber bumper dots, front and rear, to provide a little more chassis ventilation and damping. Centered and 3½ inches either side.
• Optionally add ≈0.062" spacers between top and bottom along the sides to prevent the chassis caving in on the side. My covers provide the necessary support.

See the Pudding @ lab on the cheap for the results of the above modifications.

Covers CoverDwg
Tube voltages can be fatal. The M-125s have exposed parts with lethal voltages. Tubes are hot enough to cause severe burns. These covers are simple to fabricate. Metals Depot will cut perforated sheet to size. A metal shop can bend in a few minutes. PDF of CAD dwg 0.063 material fits between chassis & base and secures nicely. Depending on hole positions, a slight corner trim maybe required to clear chassis front and rear panel overlaps.

like WOW!!!

GainComp DistComp
One of the M-125 peculiarities is the same tube for both input and phase splitter. It's not something often seen. The supplied 5963 low voltage gain of ≈20 contributes to lower than specified gain.
In a tri-amped system, it's a PITA. Something a bit higher gain like a 12AT7 or one with a bit less gain might work. A tube mentioned as a lower gain 12AT7 is the 6829. Like the 5963 it's a computer tube. Tube parameters and Load Line calculations for existing circuit component values exhibit ideal gain with voltages and currents well within 6829 specifications.
Replacing v6 with a 6829 and v7 with a 12AU7 [both NOS] increased and linearized the gain and lowered distortion! Sonically, more open and relaxed, especially when the program gets thick and robust. As always YMMV.

Familiar, decades old recordings reveal heretofore unexperienced detail and transparency, exceeding all expectations. Post listening evaluation, bench performance tests for gain and 1KHz distortion confirm the improved sonics.