Taming The Caswell #39

[Cosmo]

I've had limited success working with this design. I'd like to master it as I want to experiment with a 4 gain stage preamp. I thought the Caswell #39 would be a good place to start since there's a somewhat accurate schematic floating around on the web.

The problem is when I engage the final gain stage it becomes a hissy mess. On the scope you can see what's going on. I use a 2.5Khz wave at 500ma which is comparable to the 57 PAF HBs on my Les Paul. The amp volume and gain stage volume are both dimed.

For the third gain stage I'm using a 33K cathode resistor and a .68 bypass cap.

The signal coming out of the first gain stage (clean) is 5VAC which is fine.
The signal coming out of the second gain stage (O/D) is 10VAC without the cathode bypass cap and it's 35VAC with the cap.
The signal coming out of the third gain stage (Extreme O/D) is 65VAC without the second stage bypass cap and it's 148VAC with the cap. Both of these readings are with the third stage bypass cap.

The clean stage is fine. The second stage is fine without the bypass cap and it gets hissy with the cap. The third stage is horrible with our without the second stage bypass cap.

Obviously, I can remove the bypass caps from the two gain stages, but It's my understanding the Caswell circuit uses them. So my question is how did he get this to work?

I can post the scope traces that show all of this, but the voltages I listed above tell the story. The traces show the signal getting seriously mangled by the time it leaves the cathode follower.

Any thoughts?

[Reeltarded]

The actual amp rages with hiss when the master isn't way up with the gain moderated.. still over the top though. You have to ride the volume on the guitar in spaces or it howls like a wolf around 400hz when it sounds good, and squeals like a banshee when it sounds great. And it does sound great!

Just things to consider. The original amp is not quiet, and it's not at all smooth. Crashing is more like it.

[Structo]

Yeah most high gain amps tend to be hissy.
If you can keep the pre gain down as low as possible for the tone you want it can help.

Also if you play with the input impedance it can help for a particular guitar.

[roberto]

It's a different sound, but if you want to dampen squeals and hiss, you can modify 2nd and 3rd stages this way:

2nd stage:
100k//220p
2k2//680n

Co: 4n7
Voltage divider: 470k 470k

3rd stage:
100k//470p
10k

Co: 10n
voltage divider: 470k 470k

[John_P_WI]

A few general comments that apply to high gain.

Lowering the input grid stopper (shown as 68k) and using a higher wattage metal film here and also for the 1 meg grid leak - will lower hiss / noise.

Similarly, all of the 100k plate loads should be higher wattage packages and preferably metal films.

Also, none of the plate load resistors are bypassed with low value caps. You can try 50p. 100p, 250p etc, when the tone / feel is affected back off one value (smaller) - unless you like the effect.

There is no high frequency shunting anywhere shown and also there is minimal interstage attenuation. I think I would try to add another 470k resistor (just tack in parallel for testing) across the voltage divider to ground after the "V4 1/2 tube added" in the scheme as some attenuation, even doing the same before the tube (across the grid leak) could help. This will take gain out.

Finally, every stage has a 470k // with a 500 p "treble peak cap", you could try to raise the value of one or more to 1n, or snip a few out of the circuit.

Just some thoughts to help tame the beast.

[Cosmo]

Thanks guys. I was going to start with the treb caps on the voltage dividers. They looked like an obvious source of hiss. Also will lower the gain on the two O/D stages.

I'm running the preamp into a 6V6 SE power amp for purposes of tuning the circuit. Once I get it working properly I'll build it into a 40W 6L6 fixed bias amp. I'm going for a warm (yet crisp) BF Fender clean stage that can be relay jumped into the successive O/D gain stages.

I guess what stumps me the most is how this circuit could possibly work. I have to assume the schematic is wrong and the author never used it to build an amp. After all, I did find it on the internet so it must be correct!

[Reeltarded]

I hate to ask.. have you seen the epic SIR thread over at Metro Amp? Check the Marshall amp sub board. It's like 400 pages or some nonsense. Yikes!

Ummm the schematics you will find are all drawn by ear, so take them with a spoonful of sugar.

The #39 amp sounds very much like a Randall RG amp from the same period with a Rat, drive 0, tone 5, output max. I know this for a fact! (btw)

[M Fowler]

All the squeal and fuss is what makes the amp so much fun to play through.

I really enjoy my version of the Darkbluemurder (DBM) #39.

[Reeltarded]

Absolutely. Completely. Totally. I wish I could make any amp do that without a pedal. I can play those amps, I just can't make one.

I dream. I wish I could. I wish I just had a Randall RG sometimes. I love those plugged into tube power. Love those... Mmm..

[Roe]

check out the last schems and layouts at the metro board. I've updated mine and it works fine now

[Darkbluemurder]

The signal coming out of the first gain stage (clean) is 5VAC which is fine.
The signal coming out of the second gain stage (O/D) is 10VAC without the cathode bypass cap and it's 35VAC with the cap.
The signal coming out of the third gain stage (Extreme O/D) is 65VAC without the second stage bypass cap and it's 148VAC with the cap. Both of these readings are with the third stage bypass cap.

Obviously, I can remove the bypass caps from the two gain stages, but It's my understanding the Caswell circuit uses them. So my question is how did he get this to work?

I have done some experimenting with this circuit, and I believe my final specs are around the net. I left the 2nd and 3rd stage without a bypass cap because it made the amp shrill and caused the noise you described. Of course then the gain was no longer enough but that was the goal in the original #39 as reported by Tim Caswell - he wanted a 2203 with a bit more gain and sustain that retained the clarity of the stock model. It is true that in a later mod for George Lynch he used the bypass caps but that amp had another gain control between the 2nd and 3rd stage so it could be controlled somewhat.

To increase gain I got better results in changing the 10k cathode resistor on the second stage to 5k. And, of course, you need to remove that bypass cap on the gain control - I think this makes the amp unuseably bright at any setting below full gain.

If you want/need the bypass caps you could think about other measures to cut down the hiss - try a 500pf cap across the 1st gain stage plate resistor. That would then make it closer to #36.

Cheers Stephan

[Roe]

try hot/plate shielding, a 470pf snubber, and a 3k3 cathode instead of 10k

[Bob Simpson]

try hot/plate shielding, a 470pf snubber, and a 3k3 cathode instead of 10k

I don't understand connecting the shield to the plate.

How does this cancel noise?
Or is there some other effect which occurs?

TIA

Bob

[roberto]

Every shielded cable has a capacitance [pF/m],
usually around 100-300 pF/m, so connecting the shield to the plate
you'll increase the input capacitance of the triode, and this can dampen highs.

EG
you have 20cm of a 200pF/m shielded cable, so 200*0.2= 40pF more input capacitance, so around 25% more.

[Structo]

That is not a safe suggestion, putting the plate voltage on the shield.

You might be OK with it but down the line if a new owner gets inside might get the surprise of their life since most of the time the shield is at 0 potential.