[coinmaster]

Higher impedances will be easier loads and give less supply ripple.

 

[CopperFox]

This talk about the PSU needing more capacitance for ripple current handling is interesting as on the first page it was stated that the PSU would be kind of overkill already as it is so I've only replaced the PSU output caps with the super fast 400uf 400v DCL Wimas.

But wouldn't it be quite possible to replace the four 1000uf 200v caps with bigger ones that would be specifically designed for ripple current handling, and what would be the highest sensible capacitance that could be used here? Are there some other specific components around/in the PSU that would suffer from overcurrent with a capacitance too high?

 

[coinmaster]

Bigger is good. Lower esr is better.
I put 0.5 miliohm esr 100uf caps in there and it changed the amp completely.

But screw the caps. Get it regulated.
The amount of regulation that a simple regulator would provide is going to be vastly more than any capacitance you can realistically throw into it.
The issue is implementing it physically because you would really need a custom PCB to apply it, which is once again, getting into territory where it's easier just to redesign the entire thing.

If people want to make it worth my while, I can conjure up some PCBs that you can self populate that is the equivalent of an idealized version of the MK6.
But it would be quite out of my way so I would need some expressed interest.

 

[baronbeehive]

This talk about the PSU needing more capacitance for ripple current handling is interesting as on the first page it was stated that the PSU would be kind of overkill already as it is so I've only replaced the PSU output caps with the super fast 400uf 400v DCL Wimas.

But wouldn't it be quite possible to replace the four 1000uf 200v caps with bigger ones that would be specifically designed for ripple current handling, and what would be the highest sensible capacitance that could be used here? Are there some other specific components around/in the PSU that would suffer from overcurrent with a capacitance too high?

Get caps specifically designed for handling ripple.

As coin says regulation would be ideal but not feasible for the LD as it is, we have gone as far as we could with the decoupling caps which as you and others have found make a massive difference because of the speed and low ESR. I went one step further with the CCS to at least get constant current from the driver stage but as to its effect on sound I don't think it made a massive difference other than cleaning up the sound somewhat. IMO the optimisation that we did with, the bias points for example had greater effects on sound.

I don't really see that the PSU could suffer with too high a capacitance so long as peak demands were manageable, also the high capacity would be better able to handle any power surges such as at startup which is why you put in upgraded transformers and diodes.

The overkill was in the 1500uF, I think 1000uF stock, capacitance which was thought to be more than enough for a decent power supply to handle variations in signal. The tradeoff was good capacity and power against possible artifacts due to signal variation and such like, obviously regulation would be better though.

BTW the differences in impedance handling of the MKVI compared to the MK8 might be why you found the problem with the impedance matcher.

 

[coinmaster]

Caps designed for handling ripple are just caps with lower ESR and better heat handling. If you want high regulation with caps you need a ton of capacitance and a low enough ESR so that capacitance isn't stifled.
You could use a capacitance multiplier which would require just a single resistor and transistor added to each cap. But you would need to adjust some values in the supply to ensure there is enough voltage headroom or it will produce ugly ripple at the output.

 

[Maxx134]

Yes, I just feel that if the cathode film cap upgrade can be that much better than what is there atm it might be worth jettisoning the CCS .

No we cannot place any cathode cap on that driver tube. Sorry that was another brain fart,
But you should try your calculated bias point instead of the CCS, and find out which sound you like better.
Some aspects are tangelbles not measured, such as soundstage depth or liveliness of the tube.
The internal capacitance & other factors at play should make tube rolling more noticeable without the CCS. Not saying if it's better or worse as I have not implemented the CCS to compare myself yet.
I probably will try that soon.

This talk about the PSU needing more capacitance for ripple current handling is interesting as on the first page it was stated that the PSU would be kind of overkill already as it is so I've only replaced the PSU output caps with the super fast 400uf 400v DCL Wimas

I believe coinmaster was referring to stock implementation.

Caps designed for handling ripple are just caps with lower ESR and better heat handling. If you want high regulation with caps you need a ton of capacitance and a low enough ESR so that capacitance isn't stifled.
You could use a capacitance multiplier which would require just a single resistor and transistor added to each cap. But you would need to adjust some values in the supply to ensure there is enough voltage headroom or it will produce ugly ripple at the output.

I think the dual regulation transistors with all the extra beefed up caps, with also the increased decoupling film caps already placed in this mod thread have taken care of ripple,
But definitely the driver stage is weak.

Although the driver stage's 2 driver tubes are efficient enough with the added decoupling film caps we used, it's mostly just adequate because they are not driven optimally anyways.

So I see that the "capacitance multiplier" circuit you posted could very well have been used to mitigate most the decoupling caps & driver stage.

Unfortunately the thread owner @Redge78 was against any form of solid state devices in the design at the time.

 

[coinmaster]

You're gunna want about 10,000uf capacitance to eliminate the ripple when using SE.

 

[baronbeehive]

No we cannot place any cathode cap on that driver tube. Sorry that was another brain fart,

Haha... x2... again! We've been out of this for too long LOL . Don't know about you but I can't imagine what I was thinking. I know perfectly well I have the CCS on the driver stage anode/cathode, and the cathode bypass cap on the power stage cathode, it's not that difficult... , even without a schematic... but for some reason I assumed that the cathode bypass went on the gain stage .

 

[baronbeehive]

Following on from the work of CopperFox, and... brain farts excepting... I have a plan. I'm thinking of relaxing the restrictions that a compact version of the amp has and go with a chassis extension using heatsink plates, easily done and that would allow some of the huge film caps that according to members here should give the amp a boost in speed, and with that sharper bass, a weak point with the amp, and improving soundstage as well. So probably the 400uF Wimas for PSU which makes a ratio of around 3:1 film to lytic cap, seems reasonable. And up the power decoupling caps to about 100uF Wima. CopperFox already verified this. Which leaves cathode and WCF caps.

For cathode and WCF caps we need to take account of the cutoff on the rest of the frequency response, but also we need to have good power handling.

So for cathode caps, going from the cathode bypass calculator which is based on the mathematical formulae, probably starting at around 100uF to 500uF, figures for lytics. So for films around 50-200uF sounds about right. We will need to watch for oscillations for the higher values.

For WCF caps Maxx noted what Redge said on page 1 that half the current of the WCF push/pull design, comes from the cathode caps. Therefore the other half must come from the WCF caps? Anyway it looks like they could do with being uprated so maybe try around 1uF and possibly upwards to see the effect on the power drain, and taking account of oscillations of course here also. We may have overlooked the importance of the WCF caps before now as Maxx said.

So all of this should maximise the liveliness of the amp to a new level. I seem to remember coin, along with Redge, threw out the old chassis with his "alfresco workbench" version of the amp way back at the beginning of the thread!

How's that for a brain fart.. a huge one at that .

 

[Maxx134]

Following on from the work of CopperFox, and... brain farts excepting... I have a plan. I'm thinking of relaxing the restrictions that a compact version of the amp has and go with a chassis extension using heatsink plates, easily done and that would allow some of the huge film caps that according to members here should give the amp a boost in speed, and with that sharper bass, a weak point with the amp, and improving soundstage as well. So probably the 400uF Wimas for PSU which makes a ratio of around 3:1 film to lytic cap, seems reasonable. And up the power decoupling caps to about 100uF Wima. CopperFox already verified this. Which leaves cathode and WCF caps.

For cathode and WCF caps we need to take account of the cutoff on the rest of the frequency response, but also we need to have good power handling.

So for cathode caps, going from the cathode bypass calculator which is based on the mathematical formulae, probably starting at around 100uF to 500uF, figures for lytics. So for films around 50-200uF sounds about right. We will need to watch for oscillations for the higher values.

For WCF caps Maxx noted what Redge said on page 1 that half the current of the WCF push/pull design, comes from the cathode caps. Therefore the other half must come from the WCF caps? Anyway it looks like they could do with being uprated so maybe try around 1uF and possibly upwards to see the effect on the power drain, and taking account of oscillations of course here also. We may have overlooked the importance of the WCF caps before now as Maxx said.

So all of this should maximise the liveliness of the amp to a new level. I seem to remember coin, along with Redge, threw out the old chassis with his "alfresco workbench" version of the amp way back at the beginning of the thread!

How's that for a brain fart.. a huge one at that .

Sounds like alotta effort for unkown performance gains.
I'm not bothering with PSU, because as coin stated, the balance design cancels PSU ripple, plus I already have film caps as decoupling..
I'm only looking to change up my cathode caps and WCF caps.

 

[baronbeehive]

Sounds like alotta effort for unkown performance gains.
I'm not bothering with PSU, because as coin stated, the balance design cancels PSU ripple, plus I already have film caps as decoupling..
I'm only looking to change up my cathode caps and WCF caps.

I will let the prospect ferment in my brain for a while before doing it . If I do I might as well go the whole hog, you're probably right but might be interesting to try faster caps for PSU, I don't know. Anyway I do'nt intend to do anything more exotic as coin suggested like rebuilding the PSU for example, too much pain and maybe not enough gain, the LD already sounds pretty good to me, good enough anyway.

If the chassis is extended then I might as well fill it with something . But I agree the cathode and WCF are the interesting things to do.

 

[coinmaster]

Well, as I said. I can build a PCB for people here of the idealized version of the mk6 where all of the theoretical flaws are already solved without changing the functional design structure (It will still act as a WCF, just without the downsides).
I'd share the schematic but I won't bother unless I decide to build it. Don't want the asian lurkers stealin it off me for nothing.

But as I said, I'll need some expressed interest for me to go out of my way.
The issue is the idealized version of the input stage requires twice the number of triodes, so it would need a new chassis and wouldn't be compatible with the existing one.
I could order a custom chassis from china but again, not worth it unless enough people are interested.
Considering the B.O.M and the time I would probably need to sell it for at least $800 if I populate the board and buy all the parts. Maybe less if I can get the expected B.O.M lower.
But I would need to order in batch to keep the cost down, so a minimum of 5 people would need to be interested.

 

[baronbeehive]

Well, as I said. I can build a PCB for people here of the idealized version of the mk6 where all of the theoretical flaws are already solved without changing the functional design structure (It will still act as a WCF, just without the downsides).
I'd share the schematic but I won't bother unless I decide to build it. Don't want the asian lurkers stealin it off me for nothing.

But as I said, I'll need some expressed interest for me to go out of my way.
The issue is the idealized version of the input stage requires twice the number of triodes, so it would need a new chassis and wouldn't be compatible with the existing one.
I could order a custom chassis from china but again, not worth it unless enough people are interested.
Considering the B.O.M and the time I would probably need to sell it for at least $800 if I populate the board and buy all the parts. Maybe less if I can get the expected B.O.M lower.
But I would need to order in batch to keep the cost down, so a minimum of 5 people would need to be interested.

Appreciate it!

Don't know if enough people are hanging out around this thread so for anyone who is maybe they should get in touch if they are interested.. and if that's OK with you.

 

[coinmaster]

Yeah, seems like a decent amount of lurkers but not many participants these days.

 

[coinmaster]

Actually what I'd probably do is make the input stage and output stage separate entities in their own separate chassis connected via XLR.
No need to lock either section into a single design or purpose.
The input stage is probably one of the best possible designs for its purpose using tubes and would make a top end pre-amp or amp stage for speakers.
Better to make it modular.