Got my 6BC8's - think about UA 175A

Hi,

yesterday I got a package with some 6BC8's - quite unexpensive! - so some time I can try to build a variable Mu compressor a la Universal Audio 175A.

I hope I got the right tubes - I thought these were kind of rare, and they were suspicously cheap for a rare item ...

Expect some general variable mu questions from me in the future, and more detailed questions about the transformer ratios, overall gain etc. of the 175A.
I won't build this right now, but I'd like to start reading background info about this. I have schemos that are so-so, just ok to read component values, but not much more. Does a service manual, circuit description, adjustment procedure etc. exist on the web? I think I got the basic operation principle - it's so similar to a 1178 except for what I'd call a "tube-ota" input. But some hints at operating levels, CV range during various amounts of compression etc. would be very useful.

JH.

PS.: Also got a 6AL5 dual diode - I laughed when I opened the package,
it's the cutest littly tube I have ever seen. (;->)

year, i think they are pretty reasnobly priced. its the 6386 from fairchild etc that are stupid money.

i've just bought a load of 5670 that i'm told manley used in the later versions of their varimu.

let us know how it goes.

Something to start with: http://www.groupdiy.com/index.php?topic=249

[quote author="Michael Krusch"]Something to start with: http://www.groupdiy.com/index.php?topic=249[/quote]

Michael: I followed the link with its many links, and following most of them, I finally found this:
http://www.digital-synthologie.de/vari-mu/images/schematic.gif

Now that's obviously yours, and you will have to face my detailed questions! (;->)

For the moment, just one question: Your topology works without an interstage compressor. Can I assume that you have kept the overall gain, from the anodes of the vari gm tubes to the primary of the output transformer, the same as in the UA175?

I'm asking this beause

(a) you name various compressor models as an inspiration for your cuircuit, not just that one.

(b) As far as I can see it, the magic of these UA compressors lies in a certain combination of threshold and loop gain for each specific compression ratio. So even when I'd change the 175 circuit a bit as well, I want to keep the overall gain factor.

JH.

way too much detail,
build it an see if it works :grin:

i saw your LA2 project on your homepage. how did you do the front pannel to such a high standard, did you send off your designs?

[quote author="toby"]way too much detail,
build it an see if it works :grin:[/quote]

What, with no idea if an interstage transformer is 1:1, 1:10 or even
10:1 ? Or anything in between? (Well, I guess the 3rd one is quite unlikely.)

[quote author="toby"]i saw your LA2 project on your homepage. how did you do the front pannel to such a high standard, did you send off your designs?[/quote]

Thanks. Designed by myself, manufactured by www.schaeffer-apparatebau.de The really bad thing was the rear panel: filing the rectangular holes, and unlike the front panel, this was steel, not aluminium. 8-(


Does anybody know the part numbers of these transformers? Then I can look up the details myself ...

JH.

utc o-series plate-to...something 3:1 will work perfect here.
Pri. z about 15-30k. You need also tranny accepts small amount of dc,
'cause varymuuu VCA really _never_ perfect dc balanced )
so, something like 30k:2k, 1-5ma dc, +10...+20dbm.
Stuff like good interstage p-b:g-0-g connected "back" great here too....
OEP 2:1 or 1:6 connected back doesn't work good here due to it is not accept any dc..
Hope this helps :grin:

[quote author="ijr"]utc o-series plate-to...something 3:1 will work perfect here.
Pri. z about 15-30k. You need also tranny accepts small amount of dc,
'cause varymuuu VCA really _never_ perfect dc balanced )
so, something like 30k:2k, 1-5ma dc, +10...+20dbm.
Stuff like good interstage p-b:g-0-g connected "back" great here too....
OEP 2:1 or 1:6 connected back doesn't work good here due to it is not accept any dc..
Hope this helps :grin:[/quote]

Thanks, ijr - the hint about dc capability because of imbalance is helpful indeed. That's something I wouldn't have thought of.

Now the 3:1 ratio - is this just a guess, or do you know this for sure?
(I would have expected upward transformation instead of downward, because the termination on secondary is higher z than the driving side.
I guess from an impedance matching point of view, it wouldn't be overly critical (?), but the transformer ratio is part of the loop gain, so if the original was 1:x and I can easily get 1:y with y slightly different from x, I would at least change the gain of the following tube stage, in order to preserve the loop gain for the gain reduction, which affects time contants and threshold together with the GR ratio.
(Just explaining why I'm asking so stubbornly.)

JH.

> 6BC8's - quite unexpensive! I hope I got the right tubes - I thought these were kind of rare, and they were suspicously cheap for a rare item ...

Not at ALL rare. 6BC8 is one of many standard TV tuner tubes. Millions were made for new TVs and stocked for quick repair of failed TVs (tuner tubes work hard and die young). The TV tube factories were ahead of the game when tube-TVs became obsolete, so many thousands of 6BC8 and similar types are around. Some radios use similar types, but the few really-good radios worth keeping used better types than 6BC8. Because it is non-linear, it isn't good for most other uses, and has escaped the fashions of audiophiles. So I'm sure you are looking at a genuine 6BC8 vari-gain tube made 1950-1970, and I bet you paid $5-$8 plus shipping.

If they were hard to find, or expensive: it is a USA type and probably not at all common in Germany. Essentially equvalent types were made by German and other Euro tube companies, so no need to use USA types. There probably isn't an exact equivalent in the German types: the tube designer and the circuit designer worked together to get the best performance at the best price using the material and tooling they had, so each type is a little different even if the TV works about the same.

The 6AL5 is cute but not needed. A couple silicon diodes will work the same in any tube compressor I ever saw. Unlike say a VTVM, a compressor does not work with very low voltage rectification so the forward drop of silicon is not an issue. Use the 6AL5 if it makes you feel warmer; myself I like the reliability of sand-state devices.

As for your wider questions about tube vari-Gm compressors: it is a very complicated field, you can study it for years, and still not understand why a certain design was done a certain way. Some limiters are just cheap, or use tricks to avoid more parts, or to use parts on hand. Some of those tricks are why we like them.

The vari-gain stage plate transformer is a VERY hard thing to design, and was a popular place to cheap-out. Resistance-loaded vari-gain stages will usuallly have transient clipping in the following stage, bad in theory but potentially a good "sound".

Impedance matching is NOT an issue. In fact with triodes you usually need an impedance MIS-match; in fact you can't avoid it. The plate resistance will be maybe 7K at idle and 70K in deep GR: which do you "match" to? Opinions vary. Often you want a mis-match in deep GR: you want to reduce gain. But output power can be a problem since you are reducing tube current on strong signals, and you need enough output to drive the following stage. You can increase gain of the output section but that leads to higher noise.

Control Voltage for a tube like 6BC8 will be near-zero at zero GR, going to about -20V at 20db GR, very-roughly 1dB per Volt.

[quote author="jhaible"][quote author="toby"]way too much detail,
build it an see if it works :grin:[/quote]

What, with no idea if an interstage transformer is 1:1, 1:10 or even
10:1 ? Or anything in between? (Well, I guess the 3rd one is quite unlikely.)



i'm just using 1:1 to isolate any other stages from DC. look at all the other vari mu design also, about half of the famous ones don't used the intersatge ( altec, gates..)

like PRR said the 6bc8 was used in consumer equiptment, so its cheap . the 6386 was only used in bizare scientific test equiptment (or somthing like that) so is pretty rare. I think Jacob used a PCC 189 in his vari mu.

i've been building loads of tube stuff recently, and the bigest money saver is to used TV tube in everything. Bought a box of 300 mullard EF80 pentodes for £10!!!.

cheers for the hint of that pannel. what software did you use to do the layout? do you mind me asking how much it was?

cheers for the hint of that pannel. what software did you use to do the layout? do you mind me asking how much it was?

Click to expand...

Hi Toby,

you can download the "Frontplattendesigner" software from www.schaeffer-apparatebau.de for free. Then you can design the frontpanel on your computer, and get information about the manufacturing price with one mouseclick - without contacting the company at all. So you can try different versions and see what they would cost.

My ELOC frontpanel was EUR 81.00 + VAT and shipping, ca. EUR 100,- total.

JH.

[quote author="toby"]i'm just using 1:1 to isolate any other stages from DC. look at all the other vari mu design also, about half of the famous ones don't used the intersatge ( altec, gates..) [/quote]

Yes. I'm quite sure that capacitive coupling would work. But then again, the transformers may be part of the sound, and asuming the original transformer wasn't 1:1, the gain of the following stages should be altered in order to compensate for a different transformer, or no transformer.

[quote author="toby"]
like PRR said the 6bc8 was used in consumer equiptment, so its cheap .
[/quote]

I'm glad to hear this, really. I just ordered them from Canada, and haven't been able to test them. But their description as NOS makes sense, then. (They have obviously been re-labeled)

JH.

Hi PRR,

[quote author="PRR"]Not at ALL rare. 6BC8 is one of many standard TV tuner tubes. Millions were made for new TVs and stocked for quick repair of failed TVs (tuner tubes work hard and die young). The TV tube factories were ahead of the game when tube-TVs became obsolete, so many thousands of 6BC8 and similar types are around. Some radios use similar types, but the few really-good radios worth keeping used better types than 6BC8. Because it is non-linear, it isn't good for most other uses, and has escaped the fashions of audiophiles. So I'm sure you are looking at a genuine 6BC8 vari-gain tube made 1950-1970, and I bet you paid $5-$8 plus shipping. [/quote]

Yes. $ 4.95
And only $ 6.00 for shipping from Canada, which is an excellent service.
(from www.thetubestore.com)

The 6AL5 is cute but not needed. A couple silicon diodes will work the same in any tube compressor I ever saw. Unlike say a VTVM, a compressor does not work with very low voltage rectification so the forward drop of silicon is not an issue. Use the 6AL5 if it makes you feel warmer; myself I like the reliability of sand-state devices.

Click to expand...

Interesting. I would have thought the difference in forward drop voltage would mess up the compressor threshold, but with low currents, the voltage drop probably isn't that high. I must carefully examine this, especially how much current flow / drop voltage there is for fast attack transients. I'll probably end up at just what you'd said, that it won't make a difference. But I'll certainly learn something in the process.

As for your wider questions about tube vari-Gm compressors: it is a very complicated field, you can study it for years, and still not understand why a certain design was done a certain way. Some limiters are just cheap, or use tricks to avoid more parts, or to use parts on hand. Some of those tricks are why we like them.

Click to expand...

I have one *main* question about the vari mu concept, which has to do with headroom of the variable gain tubes, but I'll ask about this in more detail later.

The vari-gain stage plate transformer is a VERY hard thing to design, and was a popular place to cheap-out. Resistance-loaded vari-gain stages will usuallly have transient clipping in the following stage, bad in theory but potentially a good "sound".

Impedance matching is NOT an issue. In fact with triodes you usually need an impedance MIS-match; in fact you can't avoid it. The plate resistance will be maybe 7K at idle and 70K in deep GR: which do you "match" to?

Click to expand...

Well, in case of the 175A, this should be different: The anodes of the variable gm tubes work into approx. 5200 Ohm each. This impedance is in parallel to a whatsoever varying inner plate impedance of the tubes, so a rp variation of 10:1 results in a circuit impedance variation (what the transformer sees on its primary) of only 2:1.

I also suspect that these 5200 Ohms are intentionally selected to be so small, in order to give much headroom there, and maybe also to achieve a fairly constant impedance.

Opinions vary. Often you want a mis-match in deep GR: you want to reduce gain.

Click to expand...

That's what I wanted to say above; speaking of mismatch of the (outer) plate resistance to the tube's own plate resistance.
Which, funny enough, gives a transformer a good chance to work in a matched environment despite the variaton of rp.

But output power can be a problem since you are reducing tube current on strong signals, and you need enough output to drive the following stage. You can increase gain of the output section but that leads to higher noise.

Click to expand...

Yes! And for that very reason I had thought an upward transformer might be a good idea, between the intentionally low output of the variable gm stage and the quite fixed input noise of the following amplifier.

But when I look at the circuit, there's not that _much_ possibility to step up, with the secondary terminated with 2 * 20k.

So I would _expect_ some transformer ratio between 1:2 and 1:2.5 for a fairly matched operation. (SQRT(10.4k : 40k) = approx 1:2 ).
Of course it may just be mismatched, and could be anything. I just really don't know.

Control Voltage for a tube like 6BC8 will be near-zero at zero GR, going to about -20V at 20db GR, very-roughly 1dB per Volt.

Click to expand...

Thank you!

JH.

cheers for the front pannel info.

let us know how your early prototypes sound. i'm trying to do the same project but on as little budget as possible, so it would be interesting to know the difference in overall sound. well'll have to put a wav through both and compare.

i'm also using pcc189s and 5670s for compression, so it would be interesting to know what the differences are with the 6bc8.


toby

[quote author="PRR"]
The 6AL5 is cute but not needed. A couple silicon diodes will work the same in any tube compressor I ever saw. Unlike say a VTVM, a compressor does not work with very low voltage rectification so the forward drop of silicon is not an issue. [/quote]

PRR, I have just taken a closer look at the data sheet and the 175 circuit. You're perfectly right, at these low currents the voltage drop should be in the range of a silicon device.

JH.

Hi JH!
Sorry, had no time...

3:1 is suggestion based on some experiments.
I experimented with varymu's and finished with this:
http://www.geocities.com/igor_jazz/mast_comp.html
a year ago...
Tried 188/199/bc8/5670/.......
stopped on 5670 . Good triode matching, cheap;
easy to build hybrid design with it. Not the best tube, anyway.
(luv6386)!
Note, my compressor dc coupled except 10u poly capacitor before
"output volume" switch. Trimming is hell!
So, I experimented enough with different transformers on VCA's output.
In case of capacitor coupling, you get slow attack time and motorboating limitations.
In case of cheap transformer you loose transparensy in sound.
In case of dc coupling all you need ;-) is good inherent matching
between triodes and carefully designed vca.
Good opamps and some rare stuff too-
so, not less money than transformer....)
In case of transformer you need only good trafo.
Some mismatching between triodes doesn't matter here.
It will not produce drastic pumping.
(really, if you have good matching in point of, say, -2.5 v gr voltage,
it doesn't means you will have same good matching at -5v )
Be ready to find out cv breakthrough in some conditions
about -25 db "only"! (but with good matched triodes, it can be~ -54db
at 10kHz too - very hard for THAT VCA's! )
utc a-25,a-15,a-17, a-19, 0-4, 0-8,a-34, a-35.....
and a lot of same stuff can do perfect job here.
No need in trafo's ratio more than 3:1 or 4:1.
Also, 3:1 works better on removing cv breakthrough artefacts
in case of some triode mismatching than greater ones like 7:1..
Cannot explain this fact, anyway.
And 1:x cannot be here due to frequency response limitations:
say, 30k:100k trafo with freqresp. 20-20k is very hard to wound...
Anyway, you can do some experiments with different trafos...
Excuse my bad english and hope my 2 cents can help
)
Igor.

The 6AL5 is cute but not needed. A couple silicon diodes will work the same in any tube compressor I ever saw.

Click to expand...

Agree. Fully agree.
But compression in first 2-5db sounds VERY different if you replace
6al5 to 2x1n4007, say. Why????
)))
Also, it changes a compressor knee a bit.....

[quote author="ijr"]

The 6AL5 is cute but not needed. A couple silicon diodes will work the same in any tube compressor I ever saw.

Click to expand...

Agree. Fully agree.
But compression in first 2-5db sounds VERY different if you replace
6al5 to 2x1n4007, say. Why????
)))
Also, it changes a compressor knee a bit.....[/quote]


Maybe for this condition the tube diode has a _lower_ drop voltage than the silicon diode! The current thru the diodes is limited by 7 kOhm (attack pot at its fastest position), so with signals barely above the threshold there will only be tiny currents. And on the 176, at lowest compression ratio, the rectifier threshold voltage is barely 2V. (where it's 45V at 12:1)
So yes, this makes sense, now that I look at it again. (A few days ago I thought the main difference would be the tube having a _higher_ drop voltage - I never stop learning!)

More in reply to your longer post later.

JH.

Hi again Igor!

[quote author="ijr"]3:1 is suggestion based on some experiments.
I experimented with varymu's and finished with this:
http://www.geocities.com/igor_jazz/mast_comp.html
[/quote]

This looks great!

Tried 188/199/bc8/5670/.......
stopped on 5670 . Good triode matching, cheap;
easy to build hybrid design with it. Not the best tube, anyway.
(luv6386)!

Click to expand...

Interesting! I'll concentrate on this 6BC8, but maybe I'll have to buy more of them in order to find a well-matched specimen. (?) So far, I have bought five.

Note, my compressor dc coupled except 10u poly capacitor before
"output volume" switch. Trimming is hell!

Click to expand...

I can imagine that!

So, I experimented enough with different transformers on VCA's output.
In case of capacitor coupling, you get slow attack time and motorboating limitations.

Click to expand...

I noticed the munimum attack time on 175 / 176 is rather long compared to the 1176. Had no idea this was because of stability issues. My first guess was to avoid ultra-fast transients in the inevitable CV feedthru.

Good opamps and some rare stuff too-
so, not less money than transformer....)

Click to expand...

Personally I avoid expensive opamps. I don't think they are bad, but it's stunning what performance one can get from a discrete design, built from transistors worth a few Cents.

At the moment, I want to focus on tube circuits, though. I will replace the choke in the PSU with a MOSFET, and I'll use transistors for stabilizing the heater voltage, but that's it.

(I also plan to finally build a HiFi amp with tubes, but that's a different matter.)

(really, if you have good matching in point of, say, -2.5 v gr voltage,
it doesn't means you will have same good matching at -5v )
Be ready to find out cv breakthrough in some conditions
about -25 db "only"! (but with good matched triodes, it can be~ -54db
at 10kHz too - very hard for THAT VCA's! )

Click to expand...

The THAT VCAs need a strong feedback loop to work, so I see why CV rejection gets worse on higher frequencies ...

utc a-25,a-15,a-17, a-19, 0-4, 0-8,a-34, a-35.....
and a lot of same stuff can do perfect job here.
No need in trafo's ratio more than 3:1 or 4:1.
Also, 3:1 works better on removing cv breakthrough artefacts
in case of some triode mismatching than greater ones like 7:1..
Cannot explain this fact, anyway.
And 1:x cannot be here due to frequency response limitations:
say, 30k:100k trafo with freqresp. 20-20k is very hard to wound...

Click to expand...

a 30k:100k would be too high impedance anyway!
Secondary is loaded with 40k, primary is always lower than 10k
in the 175 / 176.

Why a downward transformation would help CV rejection, I don't understand. I see that using any transformer will have a big advantage in terms of CV rejection: While it can't miraculously remove mismatch of the VCA stage, it will at least prevent mismatch of the following amp stage from making it worse. It basically gives a good CMRR to the amp. (that's why you had so much work in trimming: In addition to finding the best balance in your variable GM stage, you also had to trim CMRR of your amp, I suppose.)
So - but why would the transformer ratio make any difference? If you can reduce the remaining thumping by choosing a 2:1 transformer, you could also just cut the anode resistor values of the gm stage into half. I can't see the difference, except that a 2:1 transformer gives a lower output impedance. (Half Ra values and 1:1 transformer has twice the impedance than Full Ra value and 2:1 transformer.) But does this matter? The amp's 2* 20k grid resistors aren't exactly high, anyway.

Anyway, you can do some experiments with different trafos...

Click to expand...

Well, I fear I don't really want to. Experimenting with different brands of transformers, yes - but messing with the wrong ratios and overall loop gain, no. It's so easy to build just "a" compressor, but building one that works on a "sweet spot" for all potentiometer settings is not. That's what I love about my 1176 clone: No big change in volume when I switch from 2:1 compression to 12:1 - the 1176 designers have optimised the combinations of threshold level and loop gain / compression ratio instead of leaving it to the user. I suppose they did the same thing for their tube compressors, and I want to start at this point, not find my own combinations. Once I know how the original behaves, I can start to mod it from there. But I'm not confident to find that starting point by myself.

JH.

hi jhabile,

any prototype, progress yet? or any interesting varimu based discoveries?