Rohde und Schwarz Begrenzungsverstärker ABR / U23

[Falk]

You gentlemen are great.

Thank you so much. That takes me a step further. But one after another.

The most important in vari-mu is controlling the "thumping", that is, the breakthrough of control voltage into the audio - this happens is the two tube branches are not set equally, so the controlling DC upsets their balance, punching an amount of the control signal through to the audio path (as the dynamic unbalance is generating kinda a halfway-differential-signal). To check and adjust for this, remove audio from unit's input and apply audio (e.g. 300Hz@0dBu, that's what I use) to the control grids, monitor output audio, and adjust the trimmers (Abgleich I&II) for lowest possible amount of 300hz signal breaking through to the audio output.. Note that this won't ever null completely, we're just looking for "best compromize"

Thank you very much Jacob.
Thumping is not a problem. And the integrated calibration mode is super for getting the tubes balanced. I will try to feed the grids with a clean dc and measure again.

I am a bit undecided if I like the distortion or not. It is too audible with highly dense (frequency domain) program material as synths. I have the impression that the higher voltage on the anodes of the 6al5 leads to more distortion of the feedback signal. I will recalibrate the unit after the original R+S manual and see if that makes any difference.

1.5k plate resistors will increase distortion, but it will be mostly 2nd harmonic and largely cancelled if the tubes are balanced. Of course tubes balanced for lowest thump might not be balanced for lowest distortion.

This is where 2/3 of the distortion happens. The rest is 3rd order. The issue is that most of the 2nd and 3rd order distortion happens around 500Hz to 5k and is clearly audible.

I just remembered the time I built a particular opto based tube compressor.  The opto was a series element in the feedback network of the tube amp and, when the unit was compressing, distortion went down.

I was very underwhelmed. 

I had to laugh loud. Thank you.

True enough.  It does seem a wee bit of a cruel load for those poor valves though don't ya think? 
I've only tried  1:1 Edcors as an interstage for loading a 6BC8 triode where inductance was fine, the only thing needed was a zobel on the secondary.

I do wonder if, given what Falk discovered necessary with the Edcor in this circuit, a different interstage might be better.

However, if we're talking about 2% of fairly benign distortion then I personally wouldn't overly worry.   

I agree, I will order two other interstage TXs. Any idea? I need center tapped 1:1 (10k). Any other than Carnhill VTB9071 with a <60€ price tag?

The journey goes on. Thank you everybody.
Falk

 

[Winston OBoogie]

If the transformer is the major cause of your woes, then an alternative that comes to mind is the Haufe T188  as used in the U73B..  Haufe aren't (can't?) making it anymore but I would contact Manfred and see if he's still able to supply his copies.

Scroll down the page here:

https://www.manfreddeppe.de/menu/verkauf-for-sale/


Edit:  there are details of his reverse engineering an original 188 here which might be of use in comparing to other transformers:
https://www.manfreddeppe.de/menu/diy-u73b/

 

[Falk]

If the transformer is the major cause of your woes, then an alternative that comes to mind is the Haufe T188  as used in the U73B..  Haufe aren't (can't?) making it anymore but I would contact Manfred and see if he's still able to supply his copies.

Thats funny because I have 8 of them at home for the U73b project. Thanks for this. I ll give them a try.

 

[Winston OBoogie]

Thats funny because I have 8 of them at home for the U73b project. Thanks for this. I ll give them a try.

I knew that.

  Not.   

That's serendipity for ya!  Fantastic.  I guess I might say first try it with the single 10K resistor across the top and bottom of primary as in the U73 but, you know how to play around with things to make it work in your circuit so just do your stuff and let us know.

Good luck

 

[Falk]

Good afternoon,

I spent most of the day with the compressors again, measuring, calibrating, measuring, changing tubes and so on. Good news is, that I could lower the distortion by changing the voltage on the 6al5. Its around 1% THD now where it was 2% and more before. I will test it again during work in the next weeks but I am hopeful its going to be good.

Thats pretty good news for me since I will not mess around with other transformers again. Doing so could have meant to change the overall gain staging (again).

Is there actually some regulated tube amp / compressor power supply which one could adapt for this project? It needs 250V and 102V.

I will order Mu-Metal cans next week. I will see if this will help. Overall noise is at -83dB(FS) rms now. Or -64dB(u) rms. Honestly, this could be better. For me its okay but I am sure you gentlemen would wish for it to be better. I think without shielding the audio transformers and / or using a dc heating it wont happen to get better. Main component of this noise is 50Hz hum @ -87dB(FS). The rest is below -95dB(FS).

Anyway, thanks for all your input lately. It was quite valuable.
Have a good start of the week,
Falk

 

[Heikki]

Looking at the original U23 schematic, R9 adjust the threshold voltage for tube Rö6. Adjusting voltage on Rö6 should have the same effect on distortion as adjusting the threshold pot.

I don't like that the threshold voltage for Rö6 is taken from the Rö 1 & 2 cathode circuit. I think it's going to cause distortion when the timing cap is charging. It's probably not going to show up when measuring with steady sine waves. Timing cap charging currents flowing trough the cathode resistors doesn't seem like good idea to me, but it might not cause any audible distortion and if it does it probably doesn't matter in a sound effect.

My experience with edcor XSM series transformers is that they can't be anywhere near any power transformers or they'll pick up the hum.

 

[Winston OBoogie]

Hey Falk,
I understand 100% clearly your reasons behind not wanting to try Manfred's T188 copy.
Who knows what rabbit hole or can of worms might reveal itself once you change a major component. 
If something is working OK as is, then why change right?

On the other hand, should the mid-range distortion artifact you measure still be something that you feel is an issue, at least the T188 is there for you to try without laying out money.

Heikki's comments on time constant related artifacts reminds me of sometimes being in abject fear or paralyzed uncertainty when trying to evaluate such things. 
I have no doubt these things interact, they do.  I usually just try and find the best compromise and cross fingers.

I'm sure you'll find the correct balance.

 


   

 

[Heikki]

Heikki's comments on time constant related artifacts reminds me of sometimes being in abject fear or paralyzed uncertainty when trying to evaluate such things. 

When the timing cap is charging there is basically full wave rectified signal that has no RC filtering injected to the remote-cutoff tube cathodes and that just seems like bad design to me. I think it might cause significant distortion that probably doesn't sound too pleasing. Or is there something else going on that I'm not seeing in the schematic? It is also injected to the screen grids, maybe that cancels it somewhat or maybe it makes it worse? I'm too stupid to think about it too hard and nobody else probably cares.

 

[Winston OBoogie]

...is there something else going on that I'm not seeing in the schematic? It is also injected to the screen grids, maybe that cancels it somewhat or maybe it makes it worse? I'm too stupid to think about it too hard and nobody else probably cares.

Haha, your words sound like the ones in my head and it's what I meant by paralyzed uncertainty.

Yes the threshold voltage for the diode valve R06 is derived from there but, with R9 being labeled Kompens. Spanng., isn't that saying 'voltage compensation rather than threshold adjust?  Yes, it IS the threshold voltage, but  I was thinking that it's compensating and minimizing control voltage feed through by feeding a pre-set percentage of the opposite polarity of that voltage in common mode to the cathodes and screen-grids.   

What does the manual and test procedure say, can anyone translate?

 

[Falk]

Yes the threshold voltage for the diode valve R06 is derived from there but, with R9 being labeled Kompens. Spanng., isn't that saying 'voltage compensation rather than threshold adjust?  Yes, it IS the threshold voltage, but  I was thinking that it's compensating and minimizing control voltage feed through by feeding a pre-set percentage of the opposite polarity of that voltage in common mode to the cathodes and screen-grids.   

What does the manual and test procedure say, can anyone translate?

In my circuit I am using R63 as variable threshold and R9 to match the threshold of the 2 units.
I match the units as follows which is what the manual states:

1) With no input applied I set the cathode voltage of RÖ1 and RÖ2 to 3.25V and adjust R9 to measure 7V at the cathode of RÖ6.
2) Then I adjust R5 in calibration mode so that the tubes RÖ1 and RÖ2 are balanced.
3) With +6dB(U) applied to the input I adjust R41 so that the unit works at unity gain when being set to amp mode.
4) With +16dB(u) I adjust R63 for 10dB gain reduction in compression mode.
5) I then lower the input level to +6dB(u) again and set R9 to read +5dB(u) on the output.
6) I repeat the steps 3) - 5)

When this is done both units compress almost the same (+/- 0.4dB). They can be further balanced by changing R5 on the units again slightly to match gain reduction perfectly. (+/- 0.1dB).

I have some mastering sessions later this week and will use the compressors for this. But I was listing to music through the machine on sunday for two hours and was switching them in and out of the chain and I felt for the sound to be a tad warmer / softer now. I also put in some Telefunken EF83 tubes now but did not really notice a difference to the Siemens tubes sound wise. Only thing is that the Telefunken tubes are from the same batch and the compression of the two units is perfectly synced whether linked or not. No matter if I compress 3dB or 14dB in whatever setting. They really work like a charm.

I think the wording compensation voltage derives from compensating the behavior of the different rectifier diodes. Its really perfect to get two units to play together or at least having the same regulative behavior. I can totally understand that taking the voltage from the RÖ 1&2 cathode branch is somewhat inelegantly. Maybe one could do this differently in another revision.

I ll write again later this week.
Stay safe,
Falk

 

[Heikki]

I asked spice to do some thinking for me and spice seems to agree with me. I simulated to see what happens with the cathode/screen circuit in isolation. Of course tube models aren't perfect and I just used a model of whatever small signal pentode I happen to have, but I think it's pretty close to the truth. If the tubes are in perfect spice world balance, there's no distortion. Look at the attachment to see what happens when I simulate imbalance.

 

[5v333]

Hej Falk!

If you are interested, I have two mains transformers for tube projekts up for sale .
Edcor XPWR223-240
Hammond 370CAX

But be sure about what you need before you buy so you dont end up with an expensive part that in the end wasnt right for you. (like i did..)

Im in Gothenburg.

 

[Winston OBoogie]

Thanks for the clarifications Falk, and the sim Heikki.

Given that it's easy to derive the 7v DC threshold by other benign means, I suppose the thing I had a hard time with was that the U23 designers/engineers tapped off where they did merely for convenience. 
Those guys absolutely knew what they were doing, and were also in a position where units were not being designed/built to a strict price point.
So I assumed there had to be good reason for what they did.  Maybe not?

Keep the unit(s) in perfect balance or make the distortion from imbalance  an adjustable "feature" 

 

[Heikki]

Absolutely lowest distortion might not have been the design goal, if the purpose is to prevent overmodulation of AM transmitter or sound on film variable area recording. Making fairly stable low voltage and low impedance voltage reference costs money in 1955. Decade later in U73/U83 zener diodes seem to have been used.

How much of the distortion I simulated actually shows up in a real unit is hard to say. There's already distortion when the gain is changing and adding some on top of that might be inaudible. But then again the simulated waveform looks a lot like nasty sounding frequency doubling distortion that happens in some guitar amps. In 2021 zener diodes and resistors are cheap and any builder if worried about the distortion could just put a switch that changes where the 7V threshold voltage is taken. That way it would be easy to listen if there is any difference.

 

[Winston OBoogie]

Heikki,  excuses based on facts such as historical parts availability doesn't help.
My once exalted opinion of those engineers has now fallen, 'tis tainted, has been besmirched...

 

 

[Heikki]

But did you guys know that somebody who clearly possesses high level of technical knowledge claims that Fairchild 660 is a copy of U23.

https://www.gearslutz.com/board/so-much-gear-so-little-time/1264821-fairchild-compressor-comparison-thread-hardware-only-4.html#post14013221

I think some of you might know this character from pleasant phone conversations you have had with him.

 

[gyraf]

wow, hadn't seen that post - what a load of it

Attached here, our very own CJ's interview with Rein Narma himself about the fairchild side of the subject - not much matches the dutch legitimization claims

/Jakob E.

 

[Winston OBoogie]

What a load of bollocks!  So this is the guy that got into it with Jakob and Falk?

 

[volker]

What a load of bollocks!  So this is the guy that got into it with Jakob and Falk?

Yes, seems to be...


Falk, what ratio are you using on the interstage transformer? Also, I sent you a message concerning another topic, maybe you missed it.

 

[Winston OBoogie]

Yes, seems to be...


Falk, what ratio are you using on the interstage transformer? Also, I sent you a message concerning another topic, maybe you missed it.

OK got it.  Well, I see no reason at all to invent (it is invention) a story that links the U23 to Rein Narma, the U23 stands on its own.

I'm not Falk but I believe his interstage transformer is an Edcor 10K:10  1:1