New, Cheap Dynalo...

[Pars]

^ a sigma22 could cure most, if not all, of that

[episiarch]

I have to admit that's an upgrade possibility I hadn't given a lot of thought to.

[cetoole]

^ a sigma22 could cure most, if not all, of that

Yup. Still gotta deal with the pot, but that should largely equalize things.

[episiarch]

Oh, another interesting difference. Do I understand from the above that the HA-006+ has volume-controlled preamp outs on the back? On the Gilmore Lite they're just passive loop outs. (GS-1 has both.)

[GPH]

Oh, another interesting difference. Do I understand from the above that the HA-006+ has volume-controlled preamp outs on the back? On the Gilmore Lite they're just passive loop outs. (GS-1 has both.)

Yes, the HA-006+ has volume controlled preamp outs on the back.

[Cankin]

^ a sigma22 could cure most, if not all, of that

HA-006+ is easier on wallet than most

[Pars]

^ True. A sgima might set you back almost as much as buying the HA-006+, but is a better PSU than what is used in the HA. How much difference it might make sonically? Unknown,and probably depends on a lot of things.

[Cankin]

How about some cheaper options like Tangent YJPS, or something from Twisted Pear?

There aren't much discussions about these.

[grenert]

How about some cheaper options like Tangent YJPS, or something from Twisted Pear?

There aren't much discussions about these.

The YJPS is a nice power supply, but it's only a single rail, so no-go with the Dynalo.

[Bidoux]

The YJPS is a nice power supply, but it's only a single rail, so no-go with the Dynalo.

Two of them back to back should make a symmetrical power supply. Just like Kevin Gilmore did with BH PSU.

[Pars]

The original Dynalo as well as the Dynahi supplies are tracking as well (neg. rail tracks the positive). The sigma22 does this as well. I have seen Dynalos powered by a pair of treads also.

[Icarium]

Having experienced the Gilmore Lite/GS-1/Gilmore V2/Gilmore Balanced Reference/GS-X.. I'd say power supply seems to improve performance the most with the big power supply used in the gilmore reference/balanced reference being the best.

I've heard its a toss up between which is better as a power supply (The big gilmore one versus sigma22)... I imagine both approach as good as it gets for the circuit. I've heard that the dynahi sounds better with the sigma22 than the original power supply... I'm not sure since I've never heard it with that, but I have to attest it sounds amazing with the sigma22.

[digger945]

How about some cheaper options like Tangent YJPS, or something from Twisted Pear?

There aren't much discussions about these.

Don't know how much juice the Dynalo requires, but most shunt regs don't put out very much current. It might be close.

The Placid bipolar shunt supply from TPA is not a tracking supply.

I think there is definitely a market for a properly designed/ slightly larger tracking shunt supply.

... but I have to attest it sounds amazing with the sigma22.

Indeed.

[Maniac]

Personally speaking, I don't really feel that there's a true need for a tracking power supply. Power wise, a properly done (non-tracking) power supply would not have much in the way of imbalance between the positive rail and negative rail. Secondly, with most amps, you actually won't find a lot of difference even when the power rail is slightly imbalanced, and about the only time you will know that it have an imbalanced power rail is when you open it up and measure it live with a multi-meter. Sound wise and output measurement wise, unless it is so badly imbalanced, it should actually be extremely hard for user to notice the difference even if you have the chance to A-B compare one with imbalanced rails and a precisely balanced one.

I reckon it would have to have one of the rail off by at least 10% or more to have a noticeable effect, but if any of the regulated power rail is off by 3 percent from factory setting, you'd have more problem to worry about than just sound quality. I'll have to say that I have not tested an amp with rail intensionally imbalanced, and have never really saw any +/- rail power going out of whack from all the machines that I've opened up or modded myself. (I do mod a few equipments that I use for the fun of it.)

Thanks

David

[Pars]

Remember that this is a DC coupled design, and the servos can only deal with about 50mV of offset. DC offset is dependent upon power rail balnce aomong other things. Granted, once the offset is adjusted for a particular PSU rail delta, it should be stable. Since I build my own stuff, I'll take a sigma22 FTW. The ones I've built I have tweaked the neg. rail resistor values to get the delta within 10mV or so.

[luvdunhill]

DC offset is dependent upon power rail balnce aomong other things

just curious, can you provide a justification on why this would be the case? I would have agreed with you, until I setup a test jig that disproved my theory

[Pars]

Dunno Marc, just my observations in setting up a board with two different Elpacs (with differing rail deltas), or powering one from a bench supply and changing the output voltages. Curious as to your test jig?

[luvdunhill]

well, if you're talking about differing deltas, then once the offset is zeroed, it won't fluctuate much. You'll have asymmetric clipping of course, but I don't think anyone is debating that point either. I'm more alluding to the rails tracking at different rates, let's assume no servo. SPICE is great for this, and that's what I used initially. Also, I had a few e-mails back and forth with Brian from TPA, as he was the one who initially challenged me to find a DC coupled amp, whose DC offset drifts (appreciably) with rails drifting at different rates. Again, distortion, clipping, etc. aren't a factor in this discussion, just DC offset.

I suppose it depends how much you changed the output voltage on your bench supply? I'll see what I can dig up on my computer

[Pars]

Interesting. I'll have to play around with one again sometime and see if my observations change (been a couple of years). I don't remember how much difference there was between the two Elpacs (both WM071-1950s), but changing from one to the other I needed to adjust the offset. I don't remember whether the servos could handle the change or not, though I'm usually pretty anal about offset so may have been testing with the servo opamps out.

[luvdunhill]

Think about for example a high current power amp that uses paralleled MOSFETs in push-pull configuration that manages to use a pair of CRC supplies glued together and get almost no DC offset...

[Maniac]

Well, I actually have a HA-006+ on my bench, it's about half way though the "+" upgrade and it should be good for a bit of test with imbalanced power. Lemme grab some gears and get it setup.

[Maniac]

Well... the result was quicker than I had expected... I changed -15V rail to about -10V, the output DC's bias changed 5mV when I did that. That's done without a load, which would mean that if you got something like a 32ohm load attached, it would have dropped it down to nothing. (Well, 5mV is pretty close to nothing too.) And yes, it is done with both servo removed to ensure that servo is not trying to fix any possible imbalance.

IMHO, I think the reason that the DC bias changed is not due to the fact that this is a DC coupled amp, but the effect of rail voltage on the input JFET, which have quite an effect on the output DC bias.

[luvdunhill]

Cool, thanks for trying that! that's about what I was seeing too...

[digger945]

Maybe the parts are well thought out enough to compensate a minor adjustment in rail voltage, although 5v is more than a minor adjustment I would think. Like you guys said, when one rail is altered, it is doing so for everything connected to it, not just the output devices.

[Maniac]

Just wondering, how's everyone's take on a higher end version of HA-006+ (Might just call it HA-006++)? Basically it would use better parts all around, including stepped attenuator for volume pot. I'm thinking the price would be around $450 USD, I'm beginning to test some variations of it, and hopefully will have more conclusive results soon. (All of them are not even close to proper break-in period, so I won't comment of the actual differences, but does appear promising...)

Thanks

David