How to cure JTM45/100 ghosting without altering tone?

[VintageCharlie]

Hi folks,

been long time away from the forum - as all of us - had lots of work, but managed to stick my head out lately and get back to playing guitar more.

I love my JTM45/100 clone and after trying out some amps and listening to a billion sound clips, i think this is IT for me and for the music i like. BUT the amp has ghosting, very audible at even moderate drive levels (from about 6 on the volume and upwards), especially neck pickup on Les Paul (single note runs).

I've been going through many threads on this. On plexi palace ****** has given very insightful explanations and various suggestions what can fix the issues. Here are some links to the relevant posts:

http://www.vintageamps.com/plexiboard/v ... 879#p46140
http://vintageamps.com/plexiboard/viewt ... 905#p18158
http://vintageamps.com/plexiboard/viewt ... 15#p961017
http://vintageamps.com/plexiboard/viewt ... 15#p961135

Without going into technical details that are boyound me, here is a short summary.

The ghosting on these circuits could be circumvented by:

- Increasing henries of the choke
- Increasing screen filtering
- 2K screen resistors on V4 and V5 while useing 1K screen reisitors on V6 and V7 (to counter the PI imbalance)
- bypass choke with capacitor

****** did point towards the fact that all of these solutions are Band-Aid type of fixes and that the main cause and best solution for getting rid of ghosting is found within solving the PI imbalance issue (in a proper way, where it does nt alter the tone of the amp - e.g. minimize even order harmonics, etc.)

zaphod_phil added in this post (http://vintageamps.com/plexiboard/viewt ... 30#p961985) the following: "If grid current is pushing the bias voltage off by such a big amount, doesn't that suggest that maybe 100k grid reference resistors would actually be a better idea than the usual 220k? Maybe that's what the non-band-aid cure should be. As a comparison Hiwatt power stages use 100k grid reference resistors, and you never hear of grid voltage imbalance problems with them. They also use 22k grid blocking resistors, which also helps to control grid current. So that may be worth trying in Marshall amps as well, instead of the usual 5k6."

Also Phil suggests in his post that selecting slightly unmatched power tubes could have similar effect.

Guys, please help me to find the most suited solution, if i want to keep the amp's tone as unaltered as possible, but want to get rid of the ghosting - what has worked for you, what would you suggest?
I really want to solve this, but don't wat to mess with the amp for unnecessary experiments.

Thank you very much for your input!

VintageCharlie

[Roe]

start with the grounding scheme. try the larry grounding or multi-star grouding with one star for each node on the power string. Also, form caps carefully as larry recommends

[VelvetGeorge]

Those are all the right suggestions. I'll add: reduce low frequencies in the preamp. Lows will modulate more.

george

[rgorke]

Do you have clips of your ghosting? I had an issue with buzzy mosquito distortion. Not sure if it was ghosting or blocking distortion but this old thread really helped.

viewtopic.php?f=4&t=19696&hilit=Mosquito&start=105

[Roe]

here's a useful PI calculator: https://www.ampbooks.com/mobile/amplifi ... alculator/

[rgorke]

So, Chris's "clues" are to "fix" the PI. Looking at the PI calculator...one option is to raise the 82k, another to increase the tail to 15k or 20k, right?

Having one grid resistor larger than the other is just a band-aid. Would reducing the 220k bias resistors be a band aid. That isn't fixing the PI, is it?

[Roe]

Increasing NFB also helps. Clones typically use a presence pot which reads 35% lower than the originals, resulting in reduced NFB.

[VelvetGeorge]

You would actually need to lower the 82k to get the PI outputs equal.

Every effective fix I've ever done has involved more filtering on the screens, along with one or more other changes.

george

[demonufo]

Another thing that hasn't been mentioned is increasing the current capacity of the choke. An original style choke rated at around 100mA can crap out really horribly at peak volumes in these amps.

[Roe]

100mA should be sufficient for kt66s, although el34s may draw 140mA. I got two 45/100s with kt66s (and one with el34s). One of amp clearly ghosts more than the other and the choke is the main difference between them. The 3h amp clearly ghosts a little when running flat out, whereas the 20h (RS) amp hardly ghosts at all. (Also, the el34 45/100 hardly ghosts, perhaps because I'm using 50uf on the mains)

[V8]

Roe,

Besides the ghosting is there much difference in tone or feel between the two KT66 amps? soon I hope to fire up a 45/100 which has a 3h choke installed. I do have a 20h RS , perhaps that would be a better choice with this amp.


Thanks,

David

[Roe]

David, the volume is much the same but the RS amp runs at 530v whereas the 3h amp runs at 510v (the RS amp has much more screen sag however). The RS amp sounds a little cleaner and crisper, with tighter bass. Still, the amps sound very similar until you crank them. At that point the 3h amp sounds fat, smooth, compressed and a little grainy, whereas the RS amp sounds a little thinner and brighter with more of a screaming voice that cuts better through a dense mix (two guitars etc).

[VintageCharlie]

Sorry for the late reply on my side... Guys, thanks for all the input and hints!

In my case it is clearly ghosting that i get - the distortion sounds magical. The ghosting seems also to get in the way of sustain/feedback at some points.

I did not mention - the amp already has the Larry grounding scheme - the only thing where i deviated from the vintage design. All the filter caps, except preamp were modern RIFAs.The preamp cap was properly formed. And from what i recall, we even did forming of the caps (in the amp) according to Larrys suggestions.
I might mix things up, but i think the amp runs on 490v - to be safe for modern KT66 (ValveArt).

Sounds like altering the choke would also have a noticeable effect sound wise. (3h vs 20h) That is not what i would like to achieve - i like the thickness, wooliness and openness of the amp otherwise.
Yes, forgot that one i also had stumbled over info about the current capacity of the choke affecting ghosting, in case the capacity is maxed out. What would be the alternative options for a 3h choke? (i used Marstrans 3h choke).

What effect would lowering the 82k resistor have tone-wise? And what would be an apropriate value? Is rgorke also onto something with changing the tail resistor value?

Roes hint towards the presence pot sounds interesting and promissing too - also a simple mod to do.


Thanks and best regards,

Karlis

[rgorke]

Do you have a quick clip of what your ghost notes sound like? I am not sure if what I did will help with your problem or not.

Frankly, I am not sure what one would really call my tone problem. It sounded like a thousand bees on top of my notes. From what I have learned recently, lots of issues generate in the PI. Therefore, altering values there can help.

[Roe]

I've used George's trick - reducing the 82k to e.g. 68k - with success a few times. But the calculator referenced above actually shows that 85k/98k would be better (w/o NFB). Also, there is a semi-official marshall fix to be found in the Park 1210 schematic that Steve Grindrod was behing. Grindrod's solution is to replace the 1M on V3A with a 390k, something that works well with 100k @ 4ohms NFB on a 100w amp. You do not need to this far if you use much more negative feedback (27k @ 16ohms). I'd try 820k or 680k perhaps