Overkill tube pre/EQ called Drive-1. Maintenance update.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
600:600 or 10K:10K

a tube is just a variable resistor, or a pot to the beatnik, (not weed, they call it tea)

so put a battery on a pot and turn the knob real fast, maybe 20 times a second if you are wired up like a cross country beatnik,

so now we have current moving at 20 times a second, hey, we could connect a speaker to that and dig the rebop,

if the pot has a high resistance, we will need a high voltage to get the same current to drive the speaker,

if it has a low resistance, we only need a few volts to get the same current, like a transistor,

now speakers do not like DC, so we need a way to get the battery out of the circuit,

so we use an isolation device called a transformer.

our 20 Hz AC signal will go right through the transformer, but not the DC.

that means we can drive the speaker.

if the tube has a high resistance, then we can wind the transformer with less turns on the speaker side,

this is called a step down transformer.

voltage will be stepped down, but power going in and out of the transformer will be about the same, so current will go up.

this means we can wind transformers to match the the plate resistance to the speaker load, a high plate resistance means low current so we wind less turns on the sec to get more current,

a low plate resistance like a 6L6 means our turns ratio will be lower since the tube is already pumping out 40 ma, our speaker might only need 1 amp of current,

1 amp / 40 ma = 25, so we need a 25:1 turns ratio to match the tube to the speaker.

what if we use a  a 12AX7a tube to drive the speaker?

1 amp sec speaker current/ 1 ma tube plate current = 1000, so we would need a 1000:1 transformer,
but if our plate voltage was 200 volts, now we have 1 amp of speaker current but only 200/1000 = 0.2 volts to drive the speaker, this is not enough power, .2 x 1 = 0.2 watts,

so we use a tube that can handle 200 volts on the plate and 40 ma, maybe even 400 volts on the plate and 40 ma, that = 400 x .04 = 16 watts, now were talkin,

ok so what does this have to do with 600:600 vs 10K:10K?

they are both 1:1 transformers.

well, 600 ohms at 200 volts = 200/600 = .333 amps

our tube only puts out 1 ma, not 333, so if we saddle up that 600:600 ohm transformer to our poor lil 12AX7a, we have a mis-match.

we have 200 plate volts 12AX7a /.001 ma plate current 12AX7a = 200,000 ohms,

so we need a 200,000 ohms on the pri of our transformer otherwise the tube will not have enough poop to transmit power,

so 600:600 or 10K:10K is all about matching the transformer with whatever is driving it.

propeller heads use fancy words like Source Impedance to describe these details, but it is really just volts and amps.

now when they say 600 ohms when talkin transformers, what they really mean is 600 ohms of Reactance. this just means that the ohms change with frequency.

XL=2 pi f L,  XL being our "ohms" for the transformer.

now we can see fro the formula that as frequency goes up, Reactance or "ohms" goes up, so if we are running at 10,000 hz, we will have plenty of ohms for our vacuum tube, but down that at 20 and below, the stuff hits the fan, so we need to get our inductance, "L" up there to jack up the ohms,

so a 10K transformer will have more inductance than a 600 in order to get our ohms from 600 to 10K.

how much more?

easy, since our formula is linear, inductance and ohms are the same, so we have 10,000/600 = 16.67 times as many Henries in the 10K over the 600.

Turns wise? take the root and we get a 4:1 ratio, turns of the 10K to turns of the 600.

but with less turns we get higher flux, 4 times as much, so we watch the core when building these things,
 
Thanks CJ!

I re-read that thread and I think i gleaned a little more. PRR says:

"The primary inductance is in parallel with the reflected impedance.

"Studio grade 600 ohm" winding will be somewhat above 600 ohms at the bottom of the audio band. Say 800 ohms inductive at 20Hz. So 8K at 200Hz."

So with my 600:150 and 600Ω load, at 20hz it will still look like ~800//2400 = 600Ω.

Still, primary inductance must vary somewhat from one 600:600 repeater to the next. Probably worth having a look to see.

Best,

Ben
 
As noted above frequency response/distortion might suffer if the transformer is not used with its nominal source/load impedances, though."

Pretty much the crux of the problem.  The low Z 1:1s didn't sound very good the times I tried them on bench in similar output circuits.
 
plumsolly said:
The nominal impedance, ie what separates a 10K:10K from a 600Ω:600Ω, is about recommended use, which is dictated by inductance and possibly other properties.

The design limits I described were absolute maximums based on the datasheets I have. When a manufacturer states 2k4/600, I take that literally, knowing I can always drive it with lower source impedance and into far higher input impedance. Frequency response will see negligible changes.

But if I choose to drive it with higher source than 2k4, or into lower resistance than 600 ohm I am completely on my on. There is no way of knowing exactly what will happen, but the transformer will be far from the datasheet quoted numbers, and frequency response is completely off the wack. The most common thing I've seen is a somewhat unpredictable bass cut and related phase issues.

All I can say is good luck. You are entirely in experimental territory verging outside specifications of the datasheet. I'm not saying it won't work, but it will be far from optimal. At best there will be a "happy accident".
 
yes inductance will be from 5 to 40 henries depending on which 600:600 you use,

so there is some overlapping on the 600>10K thing,

used a 600:600 on the front of an LA2a for a while, worked great with large signal inputs,
but not so great on the smaller signals, that T4b module is across the sec so it has a hard time clamping a stiff source like the 600 sec,

however it did cure the hot signal problem that overdrives the stock LA2a,

power is transmitted best when the source = the load, but not line level audio.

we don't care about lighting the streets, just the mixing board,

so once you get your tube circuit workin the way you want, you try to insert the transformer and have the tube not notice,
but there is a limit to how much inductance we can use, so we try and sample about 1/10 th the current to light up the transformer and move the signal on down the line,

 
CJ said:
yes inductance will be from 5 to 40 henries depending on which 600:600 you use,

At 20hz, A 40 henry 600:150 loaded with 600Ω should look like 2*pi*20Hz*40H // 600Ω*4 = 5024 // 2400 = 1624Ω

With 2k feeding it, at 20Hz we get a loss of 1624/(2000+1624) = 0.44 = 7dB.

With 2k feeding it, at 20kHz we get a loss of 2400/(2000+2400) =  0.54 = 5.3dB.

So 1.7db down at 20Hz referenced to 20kHz.

A 5 henry 600:150 transformer under the same conditions will give -10db at 20Hz referenced to 20kHz.

Anyways, thats the inductance/low-end part of it.
 
So I built this thing up. I ended up using both Gates transformer at leasts for now. I decided to forgo the bias switches, in part because I would have to run them out to the front panel, as I am not set up for relays. Right now I have 300Ω//470uf for V3B cathode, and a 2k rheostat with no bypass for the V1B cathode for some more distortion control. It is working well at low and middle gains, but is oscillating at the highest gains. I have found that the position and shielding of the wiring and switches for the EQ section have a large effect on the stability of the amp. I shielded all the connections to these switches and installed a metal box over the switches which helped a good bit, but I am still trying to get it stable at the highest gains. It is oscillating at frequencies between 40kHz and 150kHz, depending on where EQ wires are run and other factors. I have tried reducing high frequency gain in several places either with feedback or just a low value cap shunting signal to ground but have not been able to make it stable without eating into the audio band considerably. One angle I was curious about: I have R9-R13 mounted on the EQ switches - is there any reason that any of these resistors need to be mounted closer to the tube sockets for stability? I am considering just lowering the gain of the amp overall - I can't imagine needing 90db anyway. Let me know if you have any ideas.
Thanks,

Ben
 
I've always solved these kinds of issues by isolating gain stages. Remove all the other tubes but the stage you're diagnosing. And skip the EQ stage completely for a quick test. It's just a few rerouted wires for these isolation tests.

The EQ network is semi-high impedance and short wires do help. I had an oscillation issue on the last gain stage if you track back a few pages. But that oscillation was far higher than kHz range and very different to your interactive EQ wiring.

Also worth noting the intended Drive-1 PSU design is ultra-stiff. That will minimise lots of oscillation issues and if you are using some passive PSU that's the first place I'd look.
 
I did a little more probing today. It won't oscillate with any one of the three tubes removed. I had bypassed the eq at one point and was still able to get oscillation. I am using a passive, RLC PSU. I probed the B+ line and didn't see anything of interest when it was oscillating. For the most part, it will go into oscillation when the output attenuator is maxed and the unbypassed cathode r V1B is <800Ω or so. At settings close to oscillation, the position of the first attenuator matters - it will oscillate between about 12 o'clock and 4 o'clock of its throw. It seems to be a system-wide issue - I can't seem to isolate where the problem is occurring.

Ben
 
How did you wire the first stage attenuator? v2 schematic or older? There's a major difference.

And have you used grid stoppers anywhere? I provided a place holder for grid stoppers for every single stage on the PCB. I didn't end up needing them, but you might.

Also, just because you're not seeing oscillation directly at the B+ doesn't mean the passive PSU isn't the cause. A system wide issue nearly always points to the PSU...
 
The first stage attenuator is wired per rev 2.

I tried powering with regulated b+. No help.

I might need to explore some grid stoppers. Do you recommend a particular value?

Thanks,

Ben
 
Meh...

http://www.banzaimusic.com/910-Ohm-Carbon-Composition-0-25W.html

noninductive, good-looking, full of mojo ...

when you open your case, it will greet you ... "Hi there, I'm preventing your preamp from self-oscillation"!

it's a tech version of "inner smile"
 
The exact type and value are irrelevant.

But I'll say this, 5% tolerance on resistors on 2013 is unacceptable and 0.25 Watts is weak. Especially weak when it's more than double the size of a bog-standard 0.6watt part. Also, no article I've ever seen has shown any reliable metrics on carbon resistor "mojo", just anecdotal evidence or plain marketing bullsh*t.

Why would you still want to pay more than 10x the price of a standard part? (pro-tip, it's rhetorical)
 
I figured out one possible solution. It's not as satisfying as really figuring where the coupling is happening and eliminating it, but that may not be possible anyway. I turned the last stage into a cathode follower and that loss of gain seems to have fixed the problem.

Other options likely to have a similar effect:

Feedback - There is not a convenient place to do this. I would have to move the second stage attenuator to between V1B and V3A and take feedback from V3B plate to V3A cathode.

Putting V3A and V3B in parallel. Again, I would have to move the second stage attenuator to between V1B and V3A.

The cathode follower solution is the least intrusive of these.

Any of these three options has the dual advantage of losing gain I don't need as well as lowering the output impedance.

My concern about grid stoppers is that because my oscillation is so low (as low as 40kHz) the amount of high end attenuation I will need to fix it is going to eat into the audio band. Because of the excess of gain, I'd rather have less gain overall than any high-end roll-off. Grid stoppers might be more effective for RF oscillations?

Any thoughts on these options?

Thanks,

Ben
 
Eh, Kingston, you are ageing, showing decrease in zeal, zest and vigor ...

Of course, you can sub the mojo part for a standard carbon-film half-watter, the point is that they are (carbon-based resistance forms) sort-of "noninductive" (compared to metal-films) so will potentially introduce less of new problems when trying to cure old self-oscillation illness ...

But if you're going full-tilt, why not throw in some sexy parts?

For the bragging rights?

 
plumsolly said:
Any thoughts on these options?

I don't know if you remember, but an early revision of Drive-1 had plate to cathode feedback option between V3A and V3B. I didn't need it. Output impedance was low enough without it, and distortion at something like 0.1% already. The option is still there on the PCB actually. You can do feedback despite having the attenuator (VR4) in-between. It will be only mildly interactive. But significant enough feedback to solve your oscillation problem will take away the ability to nicely overdrive the output stage, the whole point of this project. For me at least. Then again the cathode follower does that too, with its very significant and rather nasty feedback and clipping.

Your oscillation is indeed strangely low frequency. Never encountered that before and I'm out of solutions unfortunately.

Unless you are interested in removing whole gain stage blocks. I can think of at least one way to do it well. Replace V1 with a single envelope triode, or a pentode wired as triode. You can then skip V1B stage completely and V1A could look like the NYD "two bottle" input stage. It would preserve the Drive-1 sound more like it was intended, with no feedback in sight.

[edit]

there's actually feedback on these common cathode tube stages when there's no cathode bypass cap, but that still sounds pretty smooth when overdriven.
 
As far as overdriving the output stage: As drawn the output attenuator is before V3B, so when I overdrive the unit, I am not overdriving V3B, because I have it attenuated to obtain a reasonable output level. Even with an attenuator after the output transformer, I would probably keep the level feeding V3B reasonable because I wouldn't want to overdrive the transformer. So it seems like I should't lose too much distortion capability by changing the V3B to a CF. I could set it up for maximum headroom and it would just be an interface to outside world. What do you think?

Kingston said:
Replace V1 with a single envelope triode, or a pentode wired as triode. You can then skip V1B stage completely and V1A could look like the NYD "two bottle" input stage.

I gave bypassing V1B altogether a try and it also fixed the problem, as expected. I lose my adjustable bias stage, but I can throw that on V3A and see how it sounds. Is there any reason I can't just use only half of V1A?

I think I am going to nix the plate-cathode feedback option and the parallel triode output option and experiment with the two methods above to see how they sound. Let me know if you favor the idea of one method over the other.

Ben
Kingston said:
I forgot to ask, did you implement individual B+ RC filter for each stage like in the schematic?

Yes, and I left them at 22µF.

Thanks,

Ben


 
plumsolly said:
I would probably keep the level feeding V3B reasonable because I wouldn't want to overdrive the transformer. So it seems like I should't lose too much distortion capability by changing the V3B to a CF. I could set it up for maximum headroom and it would just be an interface to outside world. What do you think?

The way Drive-1 is set up with the Cinemag as output transformer especially, it's not possible to overdrive the transformer at all. The output stage clips way before, even with the 320VDC headroom. I don't know much about your output transformer so your case might be different, but I certainly wouldn't want to lose this main method of overdrive.

plumsolly said:
Is there any reason I can't just use only half of V1A?

It's perfectly fine to use just the other half. Or you could configure the two halves in perfect parallel for mildly different THD build-up (and halved output impedance but that does nothing here).

[edit]

I should add from my experience the sound of a clipped cathode follower is the worst possible distortion, a very nasty asymmetric clip impossible to control in a meaningful way. One of the lessons learned from this project was to always avoid the possibility of clipping cathode followers in the future. Right now with VR1 into the cathode follower/filter network it's all too easy. :(
 
Back
Top