I built an amp! look for some tips

[megawhat]

ok. I tried some of the above.
I didn't liked the 220K's piggybacked. It just seemed to make the amp sound a bit different and feel different. So I left those values at 220K.

I added two 10pf caps and increase the fizz cap size.
one to ground before the 3rd stage grid, and the other to ground after the treble wiper.
That seemed to fix the worst of the high frequency oscillation without any noticeable loss of treble.

I then decided (since im not building a clone and therefore im not following strict rules) to rewire the varible cathode resistor on the CF as a fixed value (72K) using a 22K resistor and 50K linear pot. with the pot wiping feeding the TS. making it a split load output to reduce the volume post-CF.

that worked out ok! Now with no input, the hiss is LOUD when everything is at max but there is no high pitched whistle.
Ill give it more of a test run tomorrow.

[megawhat]

turns out I was wrong.
its not stable. haha

If i sit it where i work on it then it seems ok.
If it just sit the chassis on top of a cab then its not ok.
Position seems to be an issue.

going to reduce the gain in the preamp.

[paddy]

I had the same problem with an amp that I built. I fixed it by
placing a piece of sheet metal on the bottom of the chassis,
between the chassis and the bottom of the head cab. This acts
as a shield and worked a treat for me.

Cheers,
Paddy

[megawhat]

I thought about that.
I should get a head shell soon. I'll line the bottom with self adhesive copper. cheers for reminding me.

OK.
I have it a bit more stable now. Added some caps to loose treble around the 2nd and 3rd stage. Dropped a bit of gain here and there.
we'll see how it holds up for now.

[iknowjohnny]

Have you tried adding more grid blocking resistance? I have 220k's on my grids in the preamp of my hi gain affair, (10k at the input stage but 220's on then 2 following stages) and that got rid of all my issues of that type and the amp is dead stable with a good amount of gain. I would try that and remove the caps you've tried so far. i bet it will do the trick w/o killing the high end. It sure did for me.

[megawhat]

what ive got going on now isn't really killing the high end, but the grid stops should probably be increased.
I did increase them at one point and it wasn't a magic fix.
I have added a 100R grid stop to the CF.
I think I have some spare 220K's so I can try that and then start removing caps. See where (if at all) it becomes unstable again.

Thanks for all the help everyone! this is slowly getting nailed. good news it's it has pretty much put me off trying to do this again!!

[Cliff Schecht]

You shouldn't have to add any resistance to the grid of the cathode follower.. Usually just tweaking the gain, shielding the input and maybe second stage and adding grid stoppers on the first two stages and power amp is enough to kill oscillations.

With the gain, past a certain point you will notice that increasing the gain does not give a substantial increase in distortion, it just starts squealing and sounding terrible. In linear circuits you would look for the 1 dB compression point (meaning a 1 dB increase in signal doesn't give you an increase in the output signal) but here, since we are operating these devices nonlinearly, you are forced to do tweaking by ear/oscope to find this point (or by design once you know how).

[megawhat]

I don't have a scope. which is probably a mistake. but I simply don't have room for one.

I have noticed while doing tweaking (since the start of building this amp... it used to be very different) that there is leeway in reducing the level of signal at various points. I noticed I could reduce the level of the output from the cathode follower a fair bit before the tone and feel really started to change.

Reducing gain at the third stage was odd. It felt like the amp didn't clean up at all anymore, but I did experiment with quite a wide range of cathode resistor values just to see what would happen.

I tried the mesa technique of adding 120pf caps across the plate resistors for the PI but that just made my amp motorboat at a constant volume. so i took them out. bit odd!

The CF grid stop is just there because of the notes in the Merlin book... I had a spare 100R resistor and though 'why not?'.

The inputs to the first two stage are both shileded and have grid stops on. 47K on the input, but only 10K on the second stage. The third stage has no shielded wire, but does have a 10K grid stop also. I will probably increase the 10K's to 220K's and see where that leads me.

[statorvane]

I'm thinking you referenced a JCM 800 earlier in your thread. If so...

Did you try changing the bias feed resistors from 220k to 150k (like John_WP_I said)? When Marshall switched the JCM800 to use 6550s for the US market, they also changed the value of these to 150K.

Marshall also fitted a 47pF cap across the PI anodes. Might be worth trying that, even if the amp motor-boated with 150 pF.

One more thing the JCM800 had 100K plate resistors on V1 and V2 , where I noticed you have 220K for V1B and V2A. If you are looking at reducing preamp gain, that's probably a good place to start.

Hope this helps.

[megawhat]

I'm thinking you referenced a JCM 800 earlier in your thread. If so...

Did you try changing the bias feed resistors from 220k to 150k (like John_WP_I said)? When Marshall switched the JCM800 to use 6550s for the US market, they also changed the value of these to 150K.

Marshall also fitted a 47pF cap across the PI anodes. Might be worth trying that, even if the amp motor-boated with 150 pF.

One more thing the JCM800 had 100K plate resistors on V1 and V2 , where I noticed you have 220K for V1B and V2A. If you are looking at reducing preamp gain, that's probably a good place to start.

Hope this helps.

I tried piggybacking 220K's onto the 220K's already in place, but I didn't like the resultant effect on tone & feel. might be better with 150K rather than 110K total though.

I have a cap in the same place as the 47pf, that didn't cause any motorboating. The motorboating popped up when i tried to and 120pf bypass caps to the plates resistor on the PI individually.

It is JCM800-esque in that it has 3 stage and a cathode follower, and similar values around the PI. but that is all. It isn't a tweaked jcm800 clone. It started as something closer to a Soldano Atomic 16, or the crunch channel from a decatone, but with GZ34 power and a different power supply.

[statorvane]

Okay, I understand. A friend of mine had two of those amps with reverb (Astroverb I think). Tear your face off kind of grind. He's pursuing a straight-up JCM800 now. Needs more headroom for shows.

Isn't the cut control wired before the power amp grid leaks (e.g. TW Rocket)? I don't know how you have it wired, but your schematic indicates it is after the grid leaks.

Also, your presence circuit - there appears to be no cap there, and I've never seen a 250K pot in a presence circuit. I am guessing that is for a variable negative feedback, not necessarily the presence control?

Seems you are opting for a JCM800 power amp; it might be worth it to alter the preamp to match the JCM800 anyway, which is not your aim, but you could start your departures from there. Just a thought.

[megawhat]

I did go some way to making the preamp more or less the same as the JCM800 but I didn't like it that much. I like JCM800's but It started to go fizzier and fizzier and I wasn't really into that. Apart from the crazy amount of gain, it isn't that much different .

The Cut control is in the right place, and functioning. It should be at the junction of the bias feeds, power tube grid stops, and the 0.02uf caps from the PI.

The 250K is variable negative feedback. That is why the hi cut is used. Since it attenuates high frequency without being effected by that other variable NFB control. It used to have a presence and depth control. The usual style ones.

It's pretty much sorted out now. So reverting to a JCM800 would just cost me more in parts, and make my build less neat from comp swapping and stuff. I want to draw a line under it to be honest and move on to something else. haha.