Getting my feet wet with PSUD

[Saurav]

Hi,

I'm trying to come up with a power supply for the JE Labs/Angela 2A3 SE amp. I want to build the 6SL7 SRPP version, but as a stereo amp on a single chassis. Schematic:




I've tried to model the power supply in PSUD, and it checks out for monoblock construction. The photos after the schematic seem to indicate a stereo amp though, and I don't understand how this power supply can provide those voltages for a stereo amp.

http://www.angela.com/catalog/how-to/EZ.2A3.html

And the page for the non-SRPP version which describes the amp in more detail:

http://www.angela.com/catalog/how-to/SE.2A3.html

Also, here's the power supply I'm trying to work on (that will be the image at the end, as I need to upload it). Here's what I'm trying to do:

* Use a common CLC section, followed by RCRC sections for each channel. The 2A3 connects after the first RC, and the 6SL7 after the second RC.

* A 2A3 is roughly 60mA, and a 6SL7 is 2-3 mA. Since I can't think of a way to model parallel power supplies, I put in I1 which is a 63mA current sink after the CLC section, this models the other channel. I2 and I3 model the 2A3 and 6SL7 respectively. The transformer is 300V ct, and someone told me to adjust for 120/115 because Hammond transformers are designed for 115V line voltage.

Does this seem OK so far?

* When I did that with the original PS schematics, the voltages didn't end up where they should. So either I'm doing something wrong here, or this amp was designed to be a monoblock, but there are two filament transformers.

* Anyway... any comments and/or advice on my power supply? Feel free to be brutal if necessary

* I guess stepped current loads can be used to see how the circuit behaves under dynamic conditions, right? So, how much should I increase the current by? I used around 10% increase, at 3, 4, and 5 seconds for the different current sinks. Is this a reasonable figure to use? How much voltage sag is acceptable, and how quickly should it settle down? I know that ideally both the difference and the time delay should be as small as possible, but I don't have any idea of what a good ballpark figure is.

* As you can see (thought not very well), the voltage on the 2A3 seems to recover fairly quickly, but the 6SL7 voltage takes almost a second to stabilise.

That's all I have so far. Please let me know if this is mostly on the right track. And please suggest any changes you might feel appropriate. I'm trying to stay away from the GZ37 in the original schematic, because all the ones I find seem to be pretty expensive.

Thanks,
Saurav

My PS so far:

[Saurav]

And the voltage graphs (it's fuzzy because I had to change the image size to make it fit):

[arnoldc]

saurav, hi there

like you, i used PSUD to model the PSU for the JE Labs SRPP 2A3.

please allow some tolerances for the PSUD software.

if your 2A3's are connected on C3, there should be no current tap I1

look here for some ideas on stereo power supply

[Saurav]

if your 2A3's are connected on C3, there should be no current tap I1

The 2A3 for one channel connects at C3, and so I thought I1 would model the current draw of the other channel, which would split off after the CLC section. Is that incorrect? If so, could you please explain why?

Thanks,
Saurav

[Morse]

Hi Saurav;

I used PSUD II to model my JE Labs 2A3 (2001 ed) power supply, and all I did was to use 125mA as the current draw for the 2A3's (actually I cheated even more and just assigned an arbitrary 130mA or so for the whole kit and kaboodle, ignoring the decoupling caps to the input stage! ).

FWIW, using a Hammond 272JX-->RCA 5U4GB-->20uF Sprague Atom-->10H Hammond 193M-->100uF Sprague Atom I get about 345VDC to the 2A3's. With a 5AR4, it's about 30V higher IIRC. I'm only using a single "CRC" for the 2A3's, and paralleled "RC's" for the 6SL7's. If you use something other than the recommended GZ37, you will need to delay the B+ in some way. You might consider the 5AR4 - SovTek's worked fine in mine with a dropper resistor, but I had a bunch of NOS RCA 5U4GB's lying around so....

On the subject of the 120V/115V thing, there's one other thing to consider, and that's your filaments. It's been my experience that Hammonds tend to run the filaments a little hot because of this, so IMHO you should check the voltages on 'em and if they're more than a few percent high, think about dropper resistors - or if you're feeling ambitious you could always go DC and add a regulator to the filament supply.....

Good luck!!
Morse

[arnoldc]

saurav, i see - that's where the fork starts (split)

hmmm, i modeled on the PSUD like what morse did, in my case (you still have the link to my schematic do you)

morse, thank you for emphasizing my oversight of not warning saurav about that (the filaments)... and on that note, i was thinking that this scheme will always have 110V for the filament. can you please check?

Example:



(The label primary should have said, secondary)

using the 300 series of hammond wired for 220V, and a filament transformer with 110V primary...

i'll tap the filament 110V primary to the 3XX 110V primary, to have 110V always regardless of the 3XX primary voltage, which in this example is 220V.

is this sound or crazy

[EC8010]

Is this sound or crazy

Crazy, I'm afraid. The transformer does not regulate, it simply ratios voltages by its turns ratio. 220V to 110V simply means a ratio of 2:1, so if the 220V going in rises by 5%, then the 110V out will also rise by 5%.

If you are really worried about heater voltages, the best thing to do is to use regulated DC.

My experience is that PSUD tends to predict slightly low voltages (only a few percent).

[arnoldc]

Crazy, I'm afraid. The transformer does not regulate, it simply ratios voltages by its turns ratio. 220V to 110V simply means a ratio of 2:1, so if the 220V going in rises by 5%, then the 110V out will also rise by 5%.

So then if I have a constant (regulated) 220V, I get 110V on the filament transformer primary while connected to the power transformer 110V tap.

Not bad for my intentions, which is to run my amp on 220V (currently wired for 110V) even if my filament transformer primaries are 110V.

Thanks. I hope I'm not missing anything else.

saurav,

do you mean to remove the 27K and 20uF on the amp circuit (which i consider as still part of the PSU) and tap the upper 6SL7 plate directly as I3? Is that the objective of your R2 and C4 in the PSUD?

[EC8010]

Originally posted by arnoldc
I hope I'm not missing anything else.

There's no problem with using the primary of a transformer as an autotransformer, but you will have to derate its secondaries accordingly. Thus, if the heater transformer takes a load of 100VA and the total rating of the other transformer is 200VA, you will have to derate its secondaries to 100VA. Actually, since transformers are not 100% efficient, you will have to derate it to less than that.

It's a shame to %&$ on your firework, but I don't think this is a terribly good idea.

[Saurav]

Thanks for all the responses.

If you use something other than the recommended GZ37, you will need to delay the B+ in some way.

Hmm, I hadn't really thought about that. So a 5AR4 has a ramped up B+? This is the difference between directly heated vs. indirectly heated diodes, right?

so IMHO you should check the voltages on 'em and if they're more than a few percent high, think about dropper resistors

Yes, I had that at the back of my mind.

or if you're feeling ambitious you could always go DC and add a regulator to the filament supply

Probably at a later "what else can I do to this" stage.

do you mean to remove the 27K and 20uF on the amp circuit (which i consider as still part of the PSU) and tap the upper 6SL7 plate directly as I3? Is that the objective of your R2 and C4 in the PSUD?

That's what I was thinking - C4 is 20uF, and change 27K to 5K. If I do that though, I'll need to put a resistor on the plate to get the voltage down to 300, and I don't know if that will affect the "SRPP-ness" of the driver stage. If the plate needs to be connected directly to the junction of R2 and C4, then I will have to use a higher resistor value at R2 to get the voltage down, which makes that stage the same (or similar) to the original PS design. Although, 27K brings the voltage down to 250V, something around 10K seems to be what I need.

This will probably all change if I go to a 5AR4 though, right? I just substituted that in there and all voltages went up by around 30V.