Fairchild type VariMu Limiter with 6L6GCs (PRR's Idea)

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

rotheu

Well-known member
Joined
Jan 8, 2006
Messages
110
Location
CCCP
Hello everybody,

I'm relativly new to this stuff and also on the forum, so any help will be greatly appreciated.

I'm building VariMu compressor prototype, heavily based on Fairchild 660/670. I'm using a pair of 6BQ5's instead of 6V6's and (i know it's wierd :shock: ) Hammond mains power transformer (two 120V primaries in series to the plates, 50V bias tap on the secondary for the output, and 5V secondary for negative feedback (10%) :wink: , so it's basically around 5:1 ) for the control amp output and it works (gives me up to -80V DC CV). Although I don't know how it will perform compared to proper transformer.
For signal amp I'm using Edcor 600/10KCT on the input and Edcor 10KCT/600 on the output. I tried 4 pairs of 6BA6's in parallel, sounds fine, 4 6BC8's in parallel, that's OK too, but then I decided to try something different, something which is still in production. I remembered reading PRR's post about possibility of using 6L6's as VariMu tube and tried that. In triode connection, regulated +250V plate voltage, and control grids biased at -16V, and 500 Ohm balancing pot between cathodes with its brush going to the ground (cathodes are at around 6.6V), a matched pair of 6L6GC's gives 14dB of gain and 25+ dB of GR! Signal amp only draws around 50mA, signal and control amps together about 115mA. I'm using external Lamda regulated PSU.
The only problem is the amount of harmonic distortion (I know 6L6 was not designed for line level aplications). I'm running sine wave from signal generator in protools through digi001's crappy A/D/D/As to the input of the compressor and the output back to protools , and monitoring it in Waves PAZ Frequency Analyzer (not the best way to test things, but that's all I have).
With 1KHz sine wave at -20dB entering the compressor (with no GR) , output peak is at -6dB (in PAZ analyzer) , second harmonic is at -70dB.
If I change the frequency to 100Hz second and third harmonics are almost equal and at around -50dB. Lowering frequency to 20Hz increases distortion a lot (hard to see). Besides that the frequency response is strange. Transformers alone are flat from 20Hz to 20 KHz. But amp's response ref. to 1KHz is +4dB at 20Hz , +3dB at 30Hz , +1.2dB at 50Hz ,
+0.2dB at 100Hz and +0.3dB at 15KHz.

Finally my questions (tired of typing):

1.Is there any way to improve the distortion performance and frequency response ? (maybe different biasing :?: )

2.Does the VU meter in VariMu limiters show the actual amount of GR in dBs or just relative representation ?

I hooked up my GR meter like Fairchild's remote meter ( across 33 Ohm resistor and 10K pot for zeroing) but it doesn't show the amount of GR I can see in PAZ analyzer, it shows a lot more. For example, VU meter shows 5dB of GR , but change in level I can see in PAZ analizer is 0.6dB :shock: .Am I doing something wrong ?

Thank you very much.
 
hey, that sounds interesting, to me at least. maybe you can post a schematic. did you use the timing components from the fairchild?
 
Yes the time constants are the same and I pretty much used fairchild's schematic exept for the differencies I explained in my post. I don't know how to post images. If you can tell me, I will draw and post the schematic.
 
posting images is quite simple.

you can e-mail it to one of us and we post. or DIY and get a free photobucket. account...

upload the photo and then copy the IMG command onder the photo into a post and should work great. :thumb:
 
I hooked up my GR meter like Fairchild's remote meter ( across 33 Ohm resistor and 10K pot for zeroing) but it doesn't show the amount of GR I can see in PAZ analyzer, it shows a lot more. For example, VU meter shows 5dB of GR , but change in level I can see in PAZ analizer is 0.6dB .Am I doing something wrong ?

If you make the 33ohm resistor adjustable you should be able to adjust the meter so that it shows the gain reduction more accurately
.
 
2.Does the VU meter in VariMu limiters show the actual amount of GR in dBs or just relative representation ?

The meter deflection is really just a representation of the current flowing through the mu stages, and has no linear relationship to gain at all. This is why you cannot use a standard vu-meter here - you need to make up your own meter scale.

You'll probably end up with something like my G10-meter - take a closer look here:

http://www.gyraf.dk/g10/g10lg.jpg

Jakob E.
 
rotheu,
I love the innovative use of the power xfmr for the sidechain amp. Good one!

The burning question, (should you tame the THD) is..how many 6L6's do you need in parallel on each side to provide the makeup gain required when you are in deep gain reduction. How much gain with no GR are you getting from 2 6L6's?

Thats how many DB's of GR you can have and still be at unity gain when inserting the limiter.

Are you connecting the suppresor grid to the plate also? Or is it perma-connected to the cathode? I forget.

I made an innocent comment once on another forum about matching 6L6's and this is what we have wrought!:
From PRR:

> I guess like matching 6L6's

"6L6 has a bent-enough Gm curve that you could build an effective vari-Gm stage around a pair. You may not be able to get 30dB clean reduction without heroics; anyway two 6L6 is pretty heroic already (idle ~200V ~180mA). But the tubes are readily available on friday night."
 
Rotheu, was kind enough to float me a schemo of his signal amp. He can explain it better but I get to host it. :grin: IF anyone wishes to put it in the group twin account please do as I just using my photo bucket.

signal_amp.jpg


thanks rotheu :thumb: :guinness: :guinness:
 
[quote author="Michael Krusch"]Messuring current through the tubes is the wrong way. Using the control voltage is much better. You can use something simular to the 1176 meter driver.[/quote]

Yeah, as long as you dont load the 220k control line very much. I guess the 1176 circuit is hi-z.

If you do keep the current meter, put it in the cathode circuit by getting the current thru a small value resistor betweern pot wiper and ground so your meter movement isn't sitting at 250volts. =) 20 - 30 ohms should do.
 
....With 1KHz sine wave at -20dB entering the compressor (with no GR) , output peak is at -6dB (in PAZ analyzer) , second harmonic is at -70dB.
If I change the frequency to 100Hz second and third harmonics are almost equal and at around -50dB. Lowering frequency to 20Hz increases distortion a lot (hard to see). Besides that the frequency response is strange. Transformers alone are flat from 20Hz to 20 KHz. But amp's response ref. to 1KHz is +4dB at 20Hz , +3dB at 30Hz , +1.2dB at 50Hz ,
+0.2dB at 100Hz and +0.3dB at 15KHz.......

Is the output tranny rated for running dc current through it? If so, how much? If not, you could be saturating the core, causing the distortion. The tubes, by your voltage markings, are drawing ~26ma each.

If I read your schema correctly, it doesn't look like the grids have a path to ground. This may also be an isssue.

Remember, the gain due to the input xformer will be ~12dB, and the attenuation of the output xformer will be ~-12dB (gain= sq root of turns ratio.)

JDiamantis
 
pucho812,

Thanks a lot for posting the schematic! :thumb:



If you make the 33ohm resistor adjustable you should be able to adjust the meter so that it shows the gain reduction more accurately

Thank you, Rob.

I tried different values instead of 33 Ohm resistor, but it doesn't change
the way the meter reads the GR, it just changes the range for zeroing the meter, for example, the bigger the value, the higher the maximum available voltage across the meter, but the change in that voltage, when in GR, will be the same with any value of that resistor. And that's what the meter is reading, the change (dB). :wink:

The meter deflection is really just a representation of the current flowing through the mu stages, and has no linear relationship to gain at all. This is why you cannot use a standard vu-meter here - you need to make up your own meter scale.

Thank you, gyraf.

That totally makes sense to me, but I'm just wondering why most of the known VariMu limiters used standart VU meters to indicate GR? Is there any way to change the range of the readings to match the actual GR amount, using standart VU scale?


I love the innovative use of the power xfmr for the sidechain amp. Good one!

Thanks, Larrchild!

The burning question, (should you tame the THD) is..how many 6L6's do you need in parallel on each side to provide the makeup gain required when you are in deep gain reduction. How much gain with no GR are you getting from 2 6L6's?

Thats how many DB's of GR you can have and still be at unity gain when inserting the limiter.

Are you connecting the suppresor grid to the plate also? Or is it perma-connected to the cathode? I forget.

Should I tame the THD? It doesn't sound to me like pleasant tube overdrive. And I don't think it's so easy to overdrive 6L6s in this application (i think they can have power dissipation of up to 30W).
Now I have a feeling the distortion problem is not in the gain stage, it's in the output transformer! Power output of the stage is 50mA X 250V , around 12.5W (correct me if I'm wrong), the Edcor transformer is rated at 0.5W !!!!! :shock: I think BA-6A output transformer made by Sowter will be much better choice (6V6 PP output 8K:600), will give a bit more gain (has lower step-down ratio), but it's 20 times more expensive!! :sad:
So I'm thinking one pair of 6L6s is more than enough. It gives 14dB of gain (input and output transformers have the the same step-up and step-down ratios, so there is no gain and maybe 1-2dB of insertion loss). Original Fairchild 670 signal amp had 7dB of gain. Adding more tubes in parallel will not increase the voltage gain a lot, but will add a lot of power which I already have more than enough. :grin:
Supressor of 6L6 is permanently(internally) connected to the cathode.

Messuring current through the tubes is the wrong way. Using the control voltage is much better. You can use something simular to the 1176 meter driver.

Thank you, Michael.

Probably it is wrong, but it was regular practice in the "golden age of the tubes" (50-60's). GR meters were either in the plate or cathode circuitry.
Using CV will not give me the actual GR readings anyways, and don't forget, the CV in this limiter is very high(or low :? ), up to -80V.


Is the output tranny rated for running dc current through it? If so, how much? If not, you could be saturating the core, causing the distortion. The tubes, by your voltage markings, are drawing ~26ma each.

If I read your schema correctly, it doesn't look like the grids have a path to ground. This may also be an isssue.

Thank you, jdiamantis.

There is no DC in the output transformer. DC is applied to the center tap of the primary in the differential stage with matched pair of tubes, balanced cathodes and the same DC resistance of both halves of the primary. It cancels. Power handling is a different story, I wrote about that in the reply to Larrchild's post.
Yes, you read the schematic correctly. The grids have no path to the ground. They are biased at -16V and dinamically rebiased by the CV to reduct the gain. If they had fixed bias it was not ganna be limiter anymore. Just a fixed gain line amp.
 
It gives 14dB of gain
if you stay under 14DBGR, you will always have enough signal to drive the next stage properly.=) Sounds reasonable

Use a small push pull power amp xfmr for an output. Who cares if your source Z is 8 ohms these days with bridging inputs? Just use a small one.
Same as a line xfmr from 1950.

And I meant "If you should find the THD problem" So i was for fixing it=)

*Looks at watch* I have a feeling PRR is going to tell us much, shortly.
I'll just shut up.
 
Use a small push pull power amp xfmr for an output. Who cares if your source Z is 8 ohms these days with bridging inputs? Just use a small one.
Same as a line xfmr from 1950.

Thank you, Larrchild.

But don't you think, if I use, for example, 5K:16 small power output transformer (i think the turns ratio is around 17:1), I'm gonna have large drop in output level because of the very high step-down ratio ? :?:
 
You have to run the 6L6 at the highest load impedance its happy with for lowest distortion. Unfortunately, it's Rp is determined by plate area and distance, and thats some plate area!. Hmm. I said I'd shut up, just wait for PRR.=)
 
> I'm thinking one pair of 6L6s is more than enough.

I would hope so.

> the Edcor transformer is rated at 0.5W !!!!!

Uh, no. The iron at the output of the vari-Gm stage has to be BIG. Think 5W or 10W, not 0.5W. It isn't the AC, it is the DC unbalance.

> amp's response ref. to 1KHz is +4dB at 20Hz, +3dB at 30Hz, +1.2dB at 50Hz, +0.2dB at 100Hz and +0.3dB at 15KHz.

Sounds like you are in gain reduction, and your sidechain amp has lower bandwidth than the signal amp. Disable limiting when measuring frequency response.

The reason you need a Hi-Fi control amp is so the amplitude and phase-shift is flat so the control signal follows the output. If the control amp is low-Fi, extreme signals will fail to pass the control amp, will be under-limited, and your disk cutter will over-shoot into the next groove. If you are not cutting wax, your control amp does not have to be THAT good. Control amp "flaws" are part of the sound of many popular limiters, especially the LA-2 which has an awful lumpy sidechain which tends to over-limit silibants (de-essing).

> I hooked up my GR meter ...but it doesn't show the amount of GR I can see in PAZ analyzer, it shows a lot more. For example, VU meter shows 5dB of GR, but change in level I can see in PAZ analizer is 0.6dB. Am I doing something wrong?

In a perfect limiter, you can increase the input level 5dB and the output does not increase at all. A "10:1" limiter would turn a 5dB increase of input into a 0.5dB increase of output, about what you have.

You have to plot BOTH input and output levels and do the math to compute the amount of gain Reduction you are really getting.

You are really hurting yourself not having two good AC voltmeters and a DC voltmeter so you can watch everything at once. Protools can do many things; not this.

> Does the VU meter in VariMu limiters show the actual amount of GR in dBs or just relative representation?

It shows Cathode Current, 100% to 0%. If Gm were proportional to current, and load were zero or infinite, it would read Gain Reduction directly. Neither assumption is true for a tube varu-Gm, but the bad assumptions semi-cancel: a standard VU meter used as a current meter gives a good-enough representation of GR for most users. The exact dB-number is not critical in actual operation: you go for "a sound" and use the meter to get close to what has worked OK before. A few limiters did use a dedicated GR meter calibrated for actual GR of those tubes in that circuit, but the advantages of a Standard Meter which can also indicate Level and can be replaced (or remoted) on a weekend are compelling.

> 500 Ohm balancing pot between cathodes

No. Dead cathode resistance must be much less than 1/Gm. For 6L6, 1/Gm tends to be 200 ohms. Use 100 ohm or 50 ohm pot. If that won't trim, select tubes until it does balance.

> something which is still in production.

How many of these are you going to make????

There's gotta be a million NOS 6BQ7, 6BC8, 6BC6, and similar remote-cutoff tubes left in the world. If you do burn up thousands of these on a profitable product, there are Chinese and Russian toob-doods who will be happy to talk new production.

> Using the control voltage is much better.

Far more expensive (in 1947 technology), and no real benefit. Let the vari-Gm tubes also buffer the CV to the meter. If you really need precision GR numbers, mark-up the scale.

THD below 0.2% at threshold or 1% at 10dB GR is really not unexpected. If you need better, get a monolithic VCA.
 
Thank you very much, PRR!

Uh, no. The iron at the output of the vari-Gm stage has to be BIG. Think 5W or 10W, not 0.5W. It isn't the AC, it is the DC unbalance.

Now I understand, but that was what I had just to try the idea. Do you think that's the reason of very high THD ?

THD below 0.2% at threshold or 1% at 10dB GR is really not unexpected. If you need better, get a monolithic VCA.

I believe, the THD I'm getting is way higher than this at low frequencies starting at around 150Hz and down to 20Hz. The lower the frequency, the higher the THD. And that's with no GR !!! :shock: IMO it's the output trafo going crazy. What do you think ? :?:

Sounds like you are in gain reduction, and your sidechain amp has lower bandwidth than the signal amp. Disable limiting when measuring frequency response.

The measurements were taken with no GR. That's why I'm thinking the problem is in the output transformer under this conditions.

In a perfect limiter, you can increase the input level 5dB and the output does not increase at all. A "10:1" limiter would turn a 5dB increase of input into a 0.5dB increase of output, about what you have.

I wasn't increasing the input level. I was lowering the threshold and watching readings of the GR and output level meters and that was the difference.

a standard VU meter used as a current meter gives a good-enough representation of GR for most users. The exact dB-number is not critical in actual operation: you go for "a sound" and use the meter to get close to what has worked OK before.

100% agree!!! :thumb: It's ok, I'm learning! :wink:

How many of these are you going to make????

I don't know, it just feels safer ! :green:


Do you think running 6L6GCs in pentode mode will have any advantage ?
I know it will give more gain, I already tried that (about 10 dB more). I connected +250V in series with 360 Ohm resistors to the screen grids.
But I also got more noise and THD, so I went back to triode connection.
Did I do something wrong ? :?:

Thank you again, PRR! :thumb: :guinness: :sam:
 
Dead cathode resistance must be much less than 1/Gm. For 6L6, 1/Gm tends to be 200 ohms. Use 100 ohm or 50 ohm pot. If that won't trim, select tubes until it does balance.

PRR,

do you mean that the wiper of that 100 or 50 Ohm pot has to go directly to the ground, or through another resistor ? :?:

Thanks.
 
Hi,

Does anybody have any recomendation regarding the output transformer, something not very expensive, but good enoughv for this application ? :?:
 
Back
Top