I built an amp! look for some tips

[megawhat]

Hello!
so I know I'm no regular and most people hear have much more knowledge than I do, but I thought I'd get stuck in and build something after messing about with my own amps a fair bit.

It's the JCM800 2204 style 3 stages + CF into a LTP setup.
It's currently got 6550's in the output and the power is provided via a GZ34.

I used the Hi Cut you guys helped me with, and I also used a spare UF5408 in the B+ a la the Komet 60. Some of the circuit tweaks have come from looking at various schematics and just trying things out.

My circuit is probably aweful, but it sounds great (for me anyway). It doesn't sound saggy or whatever some people might associate with a valve rectified amp... it sounds pretty mean/modern. But not overly so.

The problem I've had (and I thought I'd dealt with this) is high F oscillation. I also think I have very high volume. I've got 2x 6550's idle at 65mA with 420V on the plates and a 50watt hammond OT. Yet it sounds incredibly loud!

I'm thinking this volume and the gain in the preamp is probably not helping me.

so, im after some friendly suggestions. The schematic was better and i followed a lot of ideas from Merlin's book, but since removed them for ease of tweaking more than anything else...

my thoughts are
*reduce signal level just before or after the mater vol using a voltage divider.
*increase the fizz cap
*increase the size of the grid stoppers (a lot)
*increase the 3rd stage cathode resistor to 1K8 (it is 820R currently)
*add a grid stop to the CF (between 10K - 100K)
*add some caps across plate resistors or as treble bleeds?

there is a 10pf across pins 6 & 8 on the input stage not shown on the schem.

Again, apologies if the circuit is a bit of a bonehead one. I'm very much learning as I go!

[Structo]

Well, I think you are running the bias fairly hot at about 80%.

Have you tried biasing it colder, and did it sound any better.
Some amps simply sound better colder.

As far as the parasitics, they can be a bear to track down.
Do you have access to a scope?

I see you have a very cramped chassis. The lead dress might not be optimal because of this, so having PO may be a symptom of a tight chassis.
If you really like the amp, I would suggest getting a larger, wider chassis and move the guts to that.

I believe the general consensus is you can go up to about 10K on the grid stoppers before you start losing the signal.

You might try your NFB on the 8 ohm tap for giggles to see what that brings you.
Do you not have a cap there on your NFB for presence?
Or do you just rely on the cut control for the top shimmer?

Good luck!

[megawhat]

Well, I think you are running the bias fairly hot at about 80%.

Have you tried biasing it colder, and did it sound any better.
Some amps simply sound better colder.

I used the AX84 online bias calc to set this quickly. I do have bias test points in the amp. 420V to a 6550 in that calc gives 60mA at the 60% mark.

Results: For 420 Volts
6550 (Max disapation 42 watts)
60% 70% 80% 90% 100% 110%
60 ma 70 ma 80 ma 90 ma 100 ma 110 ma
Recomended Bias point for AB1 operation is 70%

I have backed it off a bit actually.

As far as the parasitics, they can be a bear to track down.
Do you have access to a scope?

Nope

I see you have a very cramped chassis. The lead dress might not be optimal because of this, so having PO may be a symptom of a tight chassis.
If you really like the amp, I would suggest getting a larger, wider chassis and move the guts to that.

The leads are pretty short although they could be a little neater. It is still neater than a lot of production amps i've seen guts of. I have played around with moving leads using a chopstick with the amp on, but it doesn't seem to have any effect on oscillation or noise. The noise floor is actually really low. I spent a lot of time reading up on grounding before the build!

I believe the general consensus is you can go up to about 10K on the grid stoppers before you start losing the signal.

yeah, I've experimented with this in other amps. I thought I could use much larger values to roll off some high end into each stage. The SLO uses a 470K grid stop in the second stage, and I think a 220K into 4th. (I may be wrong). It thought it might be worth a try to attenuate high end throughout the preamp a little bit at a time.

You might try your NFB on the 8 ohm tap for giggles to see what that brings you.
Do you not have a cap there on your NFB for presence?
Or do you just rely on the cut control for the top shimmer?

I could indeed try the NFB on the 8ohm tap! I do not have a cap on it. its just variable NFB. I have no 'standard' presence or resonance control.
That is why I chose the hi cut, because it isn't affected by the variable NFB.

The hi cut actually kills that oscillation dead when you turn it up. I decided to try and keep the amp bright and tame that brightness with the hi-cut rather than include a presence.

I probably have a few things going on contributing to the oscillation.

[Bob-I]

The hi cut actually kills that oscillation dead when you turn it up. ....

I probably have a few things going on contributing to the oscillation.

Actually the hi-cut may just be removing the frequencies so you can't hear the oscillation.

Try grounding out the grid on each stage (except the cathode follower because there's DC voltage on that one). If the oscillation goes away, you know that the oscillation is before the point you grounded.

[megawhat]

The hi cut actually kills that oscillation dead when you turn it up. ....

I probably have a few things going on contributing to the oscillation.

Actually the hi-cut may just be removing the frequencies so you can't hear the oscillation.

Try grounding out the grid on each stage (except the cathode follower because there's DC voltage on that one). If the oscillation goes away, you know that the oscillation is before the point you grounded.

cheers!
yeah, i assumed the hi cut would just be letting the out of phase oscillation collide and cancel out. It's not fixing it, just masking it.

I'll try that grounding the grids idea.

[John_P_WI]

I would look at lowering the 6550 grid leaks from 220k to 120k or 150k.

[megawhat]

I would look at lowering the 6550 grid leaks from 220k to 120k or 150k.

cool.
It originally had EL34's in it, but I had some 6550's kicking about. The bias circuit has a pretty big range so I thought I'd try out the 6550's and just left them in.
what is the effect of lower those? might sound like a silly question.

[dynaman]

Whew, that's a tight fit. Gotta be careful with the OT wires so close to the preamp circuitry.

What does the board-mounted trimpot on the input side of the board do? Adjustable cathode or grid resistor?

What's that little cap and resistor arrangement nestled between the plate and screen supplies? Heater reference?

Where did you get the layout for that? Looks kinda familiar..

[megawhat]

The layout I drew myself from scratch using the ceriatone layouts for some marshall and matchless amps as a kind of guide. The filter cap placement across the board was dictated by the grounding scheme.

The little cap between the two is indeed for heater reference. The heater wire run along the edge of the chassis under the pots and cross the middle of the 9pin sockets (rather than wrap around) as noted on the Aiken site i think.

The inner trim pot is indeed an adjustable cathode resistor of the CF. I can't recall what it is set to. I also tried using it as a split load to drive the tone stack, which decreases output from the CF but also affects the impedance into the TS. It worked pretty well, but I like the adjustable resistor idea too. I was told a selmer used to have it in.

The OT wires run under the grounded side of the preamp wiring. The OT secondary connects to ground at the same node at the PI tail.

It's a very quiet build in terms of noise. I have found 1 grounding mistake though which I didn't notice when working on the layout.

EDIT: layout related... the chassis was predrilled and is the prototyping chassis for an actual companies builds. Therefore the power and preamp valve sockets where already in the chassis. I just added the transformers, the GZ34 and a choke to where they would fit. The chassis layout (octal at the back and 9pin at the front) also affected the layout I came up with. The layout is not ideal. especially since tweaking. This amp was very different when it was first switched on.

[Cliff Schecht]

Something I immediately notice is the amount of gain you are trying to pull out of the preamp. With the first three gain stages you already have a gain of over 100,000 which is just too much to ask out of three tube gain stages like you have them configured. Too much gain will instantly bring about not only instability but an overly sensitive input section that will clip if you fart too hard. My recommendation is to drop the plate resistor of your third stage to 100k and make the cathode 4.7k or 10k (unbypassed). This last stage does NOT need to have a huge amount of gain, the first two stages take care of this, the third stage does give you your hard clipping though and will set the tonality for the rest of the amp. I think you need to tune around in this section to remove your instability problem (if it isn't lead dress related).

[megawhat]

Cheers.
thanks for the help.
yeah it is a lot of gain. I kind of wondered how it would work out when I was going along.

The level of the signal going into that 3rd stage currently has quite an effect on tone. I stuck a few values on the bottom of the potential divider before the 3rd stage. right down to 47k. 47k sounded quite rough, but Increase to 110K sounded right for me. I stuck in 100K fixed.

I probably could increase the size of the cathode resistor there and reduce the plate.

My main worry is that (apart from the oscillation are stupidly high volume, way over giggable volume for bigger venues) is changing the tone of the amp while trying to cut that down. It sounds pretty good.

I don't want to convert it into a straight up GZ34 powered JCM800, I'd rather it kept the character it has. But like you say. Its asking too much from 3 gain stages.

[John_P_WI]

Lowering the grid leaks will lower the impedence of the grid circuit allowing more stability in the tube. (Assuming so, you don't show the rest of the bias circuit) It will also lower the drive signal to the tube, reducing some of your volume issues. You can easily tack another 220k (parallel the resistors) across the existing 220k's to try.

Also, using the boot strapped master volume, why is it grounded at the pot and at the PI ground? By using boot strapped masters you can also use a much smaller pot value.

[megawhat]

Lowering the grid leaks will lower the impedence of the grid circuit allowing more stability in the tube. (Assuming so, you don't show the rest of the bias circuit) It will also lower the drive signal to the tube, reducing some of your volume issues. You can easily tack another 220k (parallel the resistors) across the existing 220k's to try.

Also, using the boot strapped master volume, why is it grounded at the pot and at the PI ground? By using boot strapped masters you can also use a much smaller pot value.

the master volume is the same as in a 2204. the ground lug is just grounded at the ground side of the filter cap supplying the PI. The pot casing is grounded on the chassis. The 68pf cap is currently there bleeding treble to ground. It was a bit of an experiment. The ground lug on the MV is only grounded in once place.

EDIT: I'll try the 220K's in parallel for sure! thanks for that! my Bias circuit is a 50K pot with two 47uf caps. two other resistors in there also. Its basically like a wider range SLO100 bias circuit. probably used in a bunch of amps. I used a UF5408 in it. total overkill.

[John_P_WI]

OK, sorry about that, at first look it appeared like you were trying to boot strap the MV. Please ignore my MV comments...

[megawhat]

OK, sorry about that, at first look it appeared like you were trying to boot strap the MV. Please ignore my MV comments...

No problem. I thought I thought I'd missed something or my drawing was poor. (my drawing is poor! haha).

I've got a lot of stuff to try, so thanks for the input (everyone).
I've been trawling through some schematics to find some more tricks for stability also.