TAB V76 Clone - Another Take

[My3gger]

Loading tubes OT pentode with 10k pri. is very different from 50k pri. I'm suspecting Edcor since the beginning because V76 is very picky about components.

 

[klem]

nice nod to guthrie!

 

[ElSmurf]

Thanks! To make some signalboards available and a bom isnt a biggie. I wont share the PSU as this is quite on the experimental side and I dont want to do a revision soon. But just to be said. Each channel is around 1k€. This isnt a cheap build.

I'm familiar with the cost, I built an Orange86 a few years back, Guavatone's take on the REDD47. Decent components, tubes and transformers quickly add up. But I'd love to have the PCB in my backlog drawer, I've always wanted to build a V76.

 

[Falk]

Good evening, the UTMs arrived today. They sound definitely better in this position and the output pad works without influencing the sound. BUT there is still this awkward distortion. I ll trace this down with a scope asap. The distortion clearly depends on the internal level. The source level has no influence so to say. If I lower the source volume and use gain make up it has the same sound like injecting more level at less gain. So something must be wrong as both channels sound the same way bad.

After a chat with a GDIY member I we wondered what is behind @gyraf and @ruffrecords thoughts on having a resistive network on the output tx (post #14 and before) - We d appreciate if we could elaborate on this again.

Thank you,
Falk

Edit: I am just scoping through the unit. Distortion happens in the second part of the amp (V3 + V4) as mentioned earlier. If I put the feedback resistor between V4 and V3 to almost 0 ohms (full feedback) the distortion is lowest but also the output level is kind of weak. I can see the distortion before the output tx happening. G2 voltage of V4 is 125V and should be 140V.

 

[ruffrecords]

Distortion in tubes is very much level dependent. This means the stage most likely to produce the most distortion is the output stage - the last tube. The things most likely to cause excessive distortion in an output stage are:

1. Load too low - which would probably be the case with the Edcor transformer but not the UTMs (assuming they are the correct ratio)

2. Insufficient quiescent current in the output stage. What is the dc voltage across the last tubes cathode resistor and what is the resistor's value?

Also, have you measured the actual output level on the transformer secondary when distortion starts?

Cheers

Ian

 

[Falk]

Thanks for your input Ian. You led me to meassure things again and as I touched the jumper of the omitted inductor L6 with the probe the strange distortion disappeared.
So I installed C26 (and still omit C25) and its going better now. Lets see what comes next. Thank you!

Edit:
In the end I installed C25 and C26 with 100p each. The search for the sound continues but we re getting close.

 

[Falk]

Also, have you measured the actual output level on the transformer secondary when distortion starts?

I will meassure those values tomorrow. Thank you!

 

[Falk]

So next thing..
I experience the same high frequency loss as @Ilya had with the Sowter inputs transformers.
I will wrap my head around this now. Red graph is Sowter Input, orange is Edcor 10k:600r in reverse with only half of the 600r winding used to have a higher ratio.

The Edcor produces a lot of distortion but delivers more high frequencies.

 

[ruffrecords]

What is the impedance of your source for these tests?

Cheers

Ian

 

[Falk]

What is the impedance of your source for these tests?

Cheers

Ian

Its 430 ohms. Thought it was 75. I should have a DI box in between right?
These plots are left channel with the four different pads involved vs right channel channel without pad:

1: No Pad


2: 450 - 200 shunt - 450


3: 1k6 - 200 shunt - 1k6


4: 4k7 - 470 shunt - 4k7

 

[beatnik]

Could be worth trying a different input transformer and understand if the Sowter is causing the rolloff.
Perhaps even a more common 1:10 would give some insight.

Since Ilya had the same problem with Sowter, perhaps the transformer is not up to specs. Don't forget Sowter has been bought off by the Carnhill group so they now use different workers, machines, materials etc. compared to when the transformer has been developed. Who knows if they are cutting some corners..
Brian Sowter is still replying to technical questions on behalf of Carnhill so maybe he can give some advice.

 

[beatnik]

BTW I was wondering what Lundahl inductors you are using and since they are unshielded do you experience any hum pickup from the power transformers ?

 

[dmp]

You can remove the filters between v2 and v3 and you get 6dB more gain, and eliminate the broadcast frequency cuts.
C25/L6/C26 completely (replace L6 with a wire)
Increase CN1 to remove the low cut. Remove 80k RN1 and you eliminate the voltage divider that cuts your level in half.
You can add filters back if you want them later and your build is debugged.
Jensen transformers work great in V76 builds (JT-115k, JT-10k61-1M)

 

[Falk]

Could be worth trying a different input transformer and understand if the Sowter is causing the rolloff.
Perhaps even a more common 1:10 would give some insight.

Its the input tx. I tried one of Manfreds T188 and got the full spectrum. Also if I feed the audio through the direct input the spectrum response is flat.

Brian Sowter is still replying to technical questions on behalf of Carnhill so maybe he can give some advice.

I thought about writing him also. Will do!

BTW I was wondering what Lundahl inductors you are using and since they are unshielded do you experience any hum pickup from the power transformers ?

I am using 1667 15mA for the V4 and 1667 10mA for V2. They are wound on a c core with oposite windings. Like a humbucker. Lundahl says they would be really good in rejecting hum. Also I installed a mu metal shield around the power toroid. I havent jet checked the the level below -60dB FS. The left side is dead quiet. There is some hum on the right channel but I think thats a tube. I will take care of this stuff once the big issues are cleared out. I only solve one at a time.

"The LL1667 and LL1668 are anode chokes for tube amplifiers. The chokes are built with two coils and are using our own special audio C-core. The coils is made using a low capacitance coil winding technique. The two coil structure greatly reduces the risk of picking up hum caused by external magnetic fields from e.g. mains transformers. The LL1667 and LL1668 are available with different core airgaps resulting in different inductance-DC current combinations on request."

You can remove the filters between v2 and v3 and you get 6dB more gain, and eliminate the broadcast frequency cuts.
C25/L6/C26 completely (replace L6 with a wire)
Increase CN1 to remove the low cut. Remove 80k RN1 and you eliminate the voltage divider that cuts your level in half.
You can add filters back if you want them later and your build is debugged.
Jensen transformers work great in V76 builds (JT-115k, JT-10k61-1M)

I will short RN1 out and increase CN1. I have some 1:10 lying here which I will try later. If they dont do it, I will order some UTMs. Their V72 output worked great in this build here.

 

[Falk]

Here with a very small microphone transformer I had at hand. This one has a dcr ratio of 20r:150r.

Sowter 1004 (red) vs Unknown tx (orange)

 

[ruffrecords]

Highish ratio mic input transformers often have an HF response that is very dependent on the driving source impedance. All the 1:10 Sowters, Jensens and Cinemags I have tested exhibit this effect. Generally speaking they are designed to work with a 150 ohm source and in most cases their response will be flat to well past 25KHz. Drive them from a 50 ohm source and the will all exhibit a peak of several dB around 25KHz or so. Drive then from an impedance higher than150 ohms and they will exhibit the kind of HF roll off you are seeing.

I would suggest you should adjust your generator output to look like 150 ohms and repeat the tests. Presumably it is capable of 0dBu output so just make a balanced 40dB attenuator from a couple of 7K5 resistors and a 150 ohm resistor and feed the generator though it to the mic preamp.

Cheers

Ian

 

[Ilya]

Here with a very small microphone transformer I had at hand

Interestingly, the Sowter tx response is in line with what I got (the graph for low Z). This confirms my suggestions that the problem is with Sowter. Another tx that you show exhibits the response close to the one I've got with 1:10 Cinemags.
Nice to have a confirmation, thanks for posting those graphs.

I would suggest you should adjust your generator output to look like 150 ohms and repeat the tests.

I tried feeding the input with both 0R and 50R output impedance and it didn't make any difference in my case. Curious to see if anything is different with Falk unit.

 

[Winston OBoogie]

Nothing interesting.

 

[Falk]

@Ilya and @Winston OBoogie thanks for chiming in. Strongly appricated.

First of all thank you all very much for your thoughts and help. Thats such a great motivation to stick to the build until everything works. I experienced that motivation with the U23 clone and now again with the v76. Its priceless that you offer years and years of experience, knowledge and know how.

Then I have to apologize to keep your heads busy with a stupid mistake I made. When layouting the pcbs I accidentally used the EF86 footprint for the E83F.
This means what was supposed to be g1 was wired to g3 and the actual g1 was disconnected. @beatnik asked for hum and after having the spectrum issues cleared I focused on hum. Hum level was higher on the right channel. About -65dB FS peak on the right channel and about -72dB FS peak on the left. The E83F were clearly sensible when I touched them with the finger. Because of this I checked the pinouts again. After rewiring the E83F there is no trace of distortion left.

I rewired the two pins and look what we have now @10dB overall gain (minimum gain) - no traces of hum and low noise. The PSU is in the unit.
(left channel green, right channel red - the dB FS values are actually real world data)

Edit: I just checked the maximum gain of the unit which is 79dB at the moment. If I dial in the full 79dB gain and ground the input I have an output level of -50.2dB FS (in my case -43dB(U)) This means the actual self noise of the device is -122dB(U) which is more or less exactly what the TAB manual states.


I ordered UTM input transformers and two of their 550H chokes today and will test them. Maybe we can realize a cost effective V76 project this way. This unit sounds simply great now. Again. Thanks for the support.

Best,
Falk

Spectrum with Sowter input and UTM output - none of the broadcast filters installed and cap between V1/V2 and V3/V4 @0.32uF:

 

[Falk]

Short update. The UTM inputs arrive tomorrow. I had a 2 day recording session with the preamp and used it for Glynn Jones overheads with Schoeps condensers. This sounds just fine. But even better, the device worked just great with no surprises. I am really curious about the new iron now.