Reverse-engineering the TAB W95c

[vinyvamos]

I recently got hold of a broken TAB W95c module, serial number 780. Once I got the unit working after replacing a number of dried up and exploded caps, my curiosity got the better of me and I started gathering all the information that the schematic doesn't have.
Schematic can be found on this wonderful website: http://audio.kubarth.com/rundfunk/index.cgi

My intention is to design and build an EQ based on this design, under my brand Atomic Analog. With a lot of squeezing it will fit into a 500 series board, which is currently in progress on my Eagle CAD .

The last two stages are interesting, in that at Part#5 (TK42c germanium) the collector load is shared between a 3K3 resistor and a choke. The final output stage which is a funky packaged TF80/30 germanium device, has a choke in the emitter path. The choke on Q4 measures 50H @ 100Hz on my Protek LCR meter with a DCr of 4300 ohms (yep!). From voltage readings the esimated DC current is 4mA. The choke on the final stage measures 3H @ 100Hz with 135Rdc. Estimated DC current is 9mA here (EDIT: PRR debunked this so need to measure current here properly. )I have recently become familiar with choke-fed tube stages through my building of an OP6-ish preamp, but this Q4 stage is puzzling me with the parallel resistor. Some learning is needed!

I have ran tests at many frequencies with a switch to change over to a 4K3 resistor instead of this choke and the only change I see is a 2dB decrease in headroom with no choke in-circuit (+16dBu down to +14dBu). Can any of you wise folks shed some light on this?

I am having issues sourcing such a high value choke (50H) that will be small enough, and to wind my own will be rather laborious. I do intend to wind the inductors in the "präsenz" circuit, but this one will be a PITA, and for only 2dB increase in headroom I would be tempted to junk it from the design...

Sourcing the transistors is a whole other issue. The output device is on ebay, but the others are nowhere to be found. My unit has various other types fitted such as ACY32, AC160a, maybe from past repair work?? I will remove each in turn and analyze accordingly... if anyone has links to data on these TK type transistors that would be awesome!

 

[abbey road d enfer]

The last two stages are interesting, in that at Part#5 (TK42c germanium) the collector load is shared between a 3K3 resistor and a choke. The final output stage which is a funky packaged TF80/30 germanium device, has a choke in the emitter path. The choke on Q4 measures 50H @ 100Hz on my Protek LCR meter with a DCr of 4300 ohms (yep!). From voltage readings the esimated DC current is 4mA. The choke on the final stage measures 3H @ 100Hz with 135Rdc. Estimated DC current is 9mA here. I have recently become familiar with choke-fed tube stages through my building of an OP6-ish preamp, but this Q4 stage is puzzling me with the parallel resistor. Some learning is needed!

I have ran tests at many frequencies with a switch to change over to a 4K3 resistor instead of this choke and the only change I see is a 2dB decrease in headroom with no choke in-circuit (+16dBu down to +14dBu). Can any of you wise folks shed some light on this?

So far I would think you are correct in your analysis. A choke-loaded stage has potentially about twice the output voltage capability compared to a resistor-loaded one. But in this case the combination of a resistor with an  inductor mitigates the outcome; anyway the output stage adds its own limitations. It's hard to tell why the designers chose to add this resistor; maybe because the inductor alone introduces limitation to the HF response, or the resistor damps a resonance peak, or a saturation artefact... You may disconnect temporarily this resistor and see what happens...

I am having issues sourcing such a high value choke (50H) that will be small enough

I'm not surprized...

, and to wind my own will be rather laborious. I do intend to wind the inductors in the "präsenz" circuit, but this one will be a PITA, and for only 2dB increase in headroom I would be tempted to junk it from the design...

I would be tempted to say that the designers had some kind of addiction to inductors. Maybe you could regain some of this headroom by increasing teh supply voltage?

Sourcing the transistors is a whole other issue. The output device is on ebay, but the others are nowhere to be found. My unit has various other types fitted such as ACY32, AC160a, maybe from past repair work?? I will remove each in turn and analyze accordingly... if anyone has links to data on these TK type transistors that would be awesome!

I would think the design allows for some accomodation with the transistor's parameters (as any good engineered product should); indeed the input device must be a low-noise type and the output must have a suitable power dissipation, but I wouldn't hesitate replacing with equivalents instead of paying high $$ for genuine types.

 

[vinyvamos]

Thanks for your input . Ah yes I had not thought to try it without the 3K3 resistor. Maybe there is indeed a nasty resonance peak that they are killing. They were certainly not afraid of the cost of wound components that is for sure! They were building to limitations of 24V supply so this could well be the reason for persuading that little bit more headroom from the design.

 

[PRR]

> Q4 ... estimated DC current is 4mA. ...final stage ...Estimated DC current is 9mA here.

> I have a W95c on the bench, ...current draw is 140mA with relay off

(relay draw omitted)

So where is the other 127mA going? The teeny preamp?

9mA seems small for a 24V line driver.

Without knowing choke DCR, I figured 23V across 15K is 1.5mA to Base. Assume hFE is 50. Emitter current is 75mA.

Knowing choke DCR is 135r, multiply by hFE. Say 6.7K equivalent base resistance. The 15K:6.7K divider will put about 7V at base and emitter, divided by 135r makes 53mA.

If Q5 Q6 sits with 16V and 0.053A then it is a good fit for a 300 Ohm load. This allows a mild step-up to nominal 600 load, for around 16Vrms @ 600r or 425mW or +26dBm.

This is ONLY if the base is swung up to the rail and down *past* common. Q5 and the coupling network may not do that without the Q5 choke.

Back up to Q5. Choke with 4,300r, parallel to resistor of 3,300r, makes 3,300r for audio but 1,867 for DC. If we accept the 4V pencil-note across a 670r emitter resistor, DC current is 6mA. If that 6mA flowed through only the 4,300r choke it would be >25V drop, which can't happen on 24V supply. With L+R together we get 11V drop or 13V at collector. Pencil-note says 8V. AHH... I see the 510r resistor Q5c to Q6b. They are very nearly direct-coupled. Where is the DC stabilization?... ah, 82K + 5Kpot to Q4b.

I'd have to think about this a long time. I think it has been tinkered and shimmed to get the spec performance out of available parts and insufficient clean-sheet design-time.

 

[gyraf]

I am having issues sourcing such a high value choke (50H)

Neutrik NTE-10/3 (mic input transformer) is ca. 80H secondary. But I can't tell how it'd behave under DC

Jakob E.

 

[vinyvamos]

Thanks for the detailed analysis PRR . I don't know why that current estimation for the last two stages wasn't so obviously wrong to me . Thanks for that, I will do a proper measurement next time it's plugged in.

I need to read your post with pen in hand (and coffee in the other) so I can try to fully understand your calculations, as I really want to get my head around this circuit.

Neutrik NTE-10/3 (mic input transformer) is ca. 80H secondary. But I can't tell how it'd behave under DC

Jakob E.

Well it could be worth trying Jakob, and they ain't pricey . I had been looking at the henry readings of ones in my trafo junk box...

 

[vinyvamos]

If Q5 sits with 16V and 0.053A then it is a good fit for a 300 Ohm load. This allows a mild step-up to nominal 600 load, for around 16Vrms @ 600r or 425mW or +26dBm.

PRR, just to clarify, are you talking about the output transistor here, component number 6? I assume so...

4V pencil-note across a 670r emitter resistor, DC current is 6mA

. Upon inspection that resistor is actually 510R, (poor schematic scan)... so that gives 7.8mA... This gives a total of 60.8mA, which leaves about 79mA in the first four stages. Can it be so much for these stages I wonder...? 12mA in Q3, 1.15mA in Q2, and 21mA in Q1 assuming a biasing to sit the emitter at 12V.

Wait a minute, does the 5K pot set the bias of the last two stages? If so, then maybe my unit is way out of line... 

 

[abbey road d enfer]

I had been looking at the henry readings of ones in my trafo junk box...

These nominal Henry values will be reduced to shred when the winding is submitted to that much DC current. This inductor needs to be gapped (or cored with extraterrestrial alloy).

 

[PRR]

> are you talking about the output transistor here, component number 6?

Sorry. Yes. Corrected.

> The choke on Q4 measures ...

I believe this should be Q5.

 

[PRR]

> maybe my unit is way out of line...

Do you have about 8V at Q6 E? That's penciled on the plan and makes sense. If it has gone sour and risen to 15V then your current will be about double and your up-swing very poor.

> does the 5K pot set the bias of the last two stages?

You would think.

But feedback from C5 to B4 is +positive+ feedback, right?

I worked out what I think is the DC NFB path via 82K and pot. On static analysis it works out fine: pot set about half-way balances the loop. However loop-closed, I think it throws the bias to one extreme or the other?

Is there even any DC gain? Q4 works at DC gain near 4. Q5 has DC gain near 3. Forward gain of 12. 85K to ~2K in pot is feedback ratio like 40:1. There's nearly no DC NFB. The pot network takes a somewhat stable arbitrary DC voltage and divides it down to give Vbe(4) plus V(300r).

FWIW: the AC gain of Q4 is a bit under unity (Q5b is a heavy AC load), Q5 gives a AC gain near 400, Q6 is unity plus any transformer step-up. Since the stage closed-loop gain is 3 (plus whatever boost, maybe to 10), there is plenty of audio NFB.

I think it has been patted and molded to just-barely meet some spec (possibly including keeping the coil-winders busy). It could be radically cheaper if you accept "only" +15dBm output. And as a practical thing, _I_ would sneak a NE5532 in at Q1 and Q456 (with optional TF80/30 if available). The Q2 Q3 stage has odd details (DC through pot tracks) but should be easy to implement.

 

[PRR]

.

 

[Nickuk99]

Hi, can you tell me how this project is progressing?

I have a pair of TAB W95c they are interesting as they have been modified for mastering use in the early 80ies according to the design of the leading engineer, Mr. Schütte, of TAB in Wuppertal, Germany.

The mid band was originally designed for boost only. It has been modified for cut only. The high and low shelf filters have been modified to extended center frequencies at 25 Hz and 16 kHz.

Specifications:

Active Bands: 3

Hi-Shelf:  16 kHz, +/- 12 dB

Lo-Shelf:  25 Hz, +/- 12 dB

Presence Cut: 0,7/1,0/1,4/2,0/2,8 kHz, - 8 dB

I look forward to hearing from you, Many thanks,

Nick

 

[vinyvamos]

Hi Nick,
I guess better late than never (3 yrs!?) But yes this project is going well and I am back working on it now after a long break. Second revision of PCBs came back last week and I am getting a small 50H gapped choke custom made too. 5H choke is obtainable from Hammond and I have a few options for in and out trafos. I am very interested to hear how your units were modified for mid cut as I am trying to add a mid cut/boost switch on mine. I have already added a switch to select 3 HF frequencies. Funny that PRR mentioned "DC on pot tracks" as what is missing from the schematic is a 50uF cap that I found in three original units that I've worked on, sitting at the base of Q2. And sure enough on my rev1 the pots cause scratch noises.