The ultimate DIY? A Stax SRM-T2!

[kevin gilmore]

You better look at the -15 again... If you put in a 12 volt regulator, then you forgot to flip the one diode.

Not that it probably matters, but if the diode is in wrong, then its gong to be about -18.

[luvdunhill]

yup, I put in a -12v regulator. I measured it again and it was -13.8V and the diode is flipped (with regard to the silk) on the -12v part (which is closest to the edge of the board). The part I used is 511-L7912CV. Regardless, the -12v regulator I got from ST is the annoying (new?) single gauge type with the paper thin tab. Since these devices aren't putting much out, I'm thinking about using plastic parts here as well (511-L7912CP and 511-L7812CP)... strange though, first time I've ever used ST regulators... once I get a connector installed there, I'll put a load on it and see what happens.

oh and the other half of the board works as well

[luvdunhill]

ST micro

[Craig Sawyers]

I

[luvdunhill]

Just thinking about this again, if the 10m90s are just going into current limit (depends on how you read the specs - somewhere between 40mA and 50mA for 30 ohms, times two of course) that could explain the oscillation. Try dropping R24 and R31 a tad - say 27 ohms or 25 ohms just by putting a temporary resistor in parallel and see if that cures it.

I think that your measured current on the -500V is about right - so when I load up my power supply (soon!) I'll look carefully for that oscillation

perhaps more capacitance is needed on the amp side to swamp out the effect of the umbilical cable?

[kevin gilmore]

A little too much inductance in the cable could possibly do this. I kept mine rather short because

i was running out of 10 colors of wire. Eventually i was going to make a much longer pair.

I just do not understand how ST can make crap like the 7912. I finally got a couple of these, and no

matter what i do, even in the most simple of circuits with 100ma of load, they just don't fucking work.

[luvdunhill]

matter what i do, even in the most simple of circuits with 100ma of load, they just don't fucking work.

they do in fact burn, given enough frustration and a "evil slip" of the bench supply. I can confirm this.

[Inu]

Just thinking about this again, if the 10m90s are just going into current limit (depends on how you read the specs - somewhere between 40mA and 50mA for 30 ohms, times two of course) that could explain the oscillation. Try dropping R24 and R31 a tad - say 27 ohms or 25 ohms just by putting a temporary resistor in parallel and see if that cures it.

I think that your measured current on the -500V is about right - so when I load up my power supply (soon!) I'll look carefully for that oscillation

Thank you for your advice.

The noise (ripple) that I am trying to reduce is generating in the NFB loop.

I got a same result on the LTspice so I will conclude this issue.

I tried to use several OP amps OP27(TI), LF351(NS), OP07(AD), OPA177(AD), so far OP27(TI) is the best on my PS.

[kevin gilmore]

looks like i'm going to have to learn ltspice...

wonder what happens if i put in something fast.

I'll try that when i feel a bit better... stomach flu and high voltage

probably don't mix.

[Craig Sawyers]

Thank you for your advice.

The noise (ripple) that I am trying to reduce is generating in the NFB loop.

I got a same result on the LTspice so I will conclude this issue.

I tried to use several OP amps OP27(TI), LF351(NS), OP07(AD), OPA177(AD), so far OP27(TI) is the best on my PS.

Interesting - mine are Analog Devices OP27GPZ. Waiting for bolts to strap down the mains transformers, then I'll fire it up into dummy loads.

[digger945]

looks like i'm going to have to learn ltspice...

I am trying to learn it myself. Kinda frustrating until you get the hang of it, then it's a breeze. I haven't found a lot of good help online, but then again I am a slow learner anyway, and not quite as quick to pick up on things as some of the younger crowd.

[kevin gilmore]

i'm having no luck... don't know what to stick (the spice model for the 2sc3675) and where to stick it, or

how to get it to simulate what Inu see's...

Then i'll get around to figuring out how to add/chance component values...

Time for an early dinner.

[Dusty Chalk]

... don't know what to stick...and where to stick it...

QFtehLulz.

You must be really off your game, go get some rest, drink plenty of fluids, and have some chicken soup. Er...not necessarily in that order (save the 'rest' for last).

[luvdunhill]

The worst thing about LTSpice is the schematic capture. Can your CAD program export a netlist? It might be easier to work with it in text form (this is what I do) and bypass the schematic capture program. If it helps, I can upload you a high voltage regulator circuit that I put together (and subsequently made boards, so yeah it works) to play with.

[Kerry]

I agree. I use DipTrace and put the spice info into each component export the net list and then load and run in LTSpice. I was able to simulate the Blue Hawaii like this. The only issue was getting the 2SJ79 to simulate correctly. I got it close in the end. A nice benefit of this is that the schematic is tied to the board layout and tied to the spice model so you know that you have things connected correctly.

[manaox2]

So, quick question, hope I'm not out of line here. Is KG being banned going to hamper developments on the T2 and KGSSHV? I have no details on reason for or length of the ban; that is not my business. I would however really really like to see these projects continue in the community.

[Dusty Chalk]

I'll let him answer that one himself when he gets back.

[spritzer]

Note from KG:

8 chassis ARE at the polisher, when they get back 4 will go for anodizing.

Then i'll ship the 4, people can contact me via email if they want the chassis. Then do the other 4. Then ship off the last 4 for polishing.

If people don't want the chassis, i'll do a corrected board run, and i know lots of people desperate for them.

fix for the power supply will be posted.

[n_maher]

Kevin can wait to supply other answers until he's allowed to post here again.

[wink]

High Quality Audio & Industrial Attenuators

any comments...?

[luvdunhill]

High Quality Audio & Industrial Attenuators

any comments...?

There's a thread on HF about them. Too low brow for this thread unfortunately...

[wink]

I found that they have problems with not supplying the goods after payment.

[wink]

Seems like supply issue is resolved...

[Les_Garten]

Seems like supply issue is resolved...

One or two deliveries after lots of non deliveries doesn't necessarily ring as resolved.

I always question the reliability of a company that won't answer "pre-sales" questions.

[kevin gilmore]

First things first. Then i will read upward later.

The power supply problem is fixed. Verified by Inu. Took about 5 tries. Will have pictures soon.

Power supply total noise is 26 millivolts at load. Under the noise of the tubes. Which is still very quiet.

schematic updated with the fix. For those that want to play...

http://gilmore.chem.northwestern.edu/T2PSfix3.asc

get a copy of ltspice which is free. Simulate and put the probe on the output.

Cut one of the wires to C3 (but not C3) and simulate again. ltspice likes to rearrange nodes

sometimes for no apparent reason, so if things look screwy, delete the trace and put the probe back on.

I still cannot get my spice machine which has never failed in the past to do what this does.

But with the added parts, the optimization still comes out the same.

I can't play any more for 2 days due to the labor day power shutdown.

Lt spice completely fails with the same spice models on the amplifier schematic.

8 chassis have been delivered to the polisher. I expect them to be back in about 3 more weeks,

then 4 go for anodizing. This is strictly a logistics problem now. All of the machining of every

single part (except for more stax jacks because i keep selling those on the side) is finished

before the increase in price of machine time on sept 1.

When the chassis arrive from the polisher i will ask for payment on the next 4 chassis.

Many who missed out on this want another run.

Justin has a brand new heatsink that is 5 inches high. (like this thing isn't big enough already)

the 5 inches combined with the 16.2 inch length of the heatsink means that i cannot machine

the front and back holes. I would like to find a machine shop that would do the entire job.

If someone knows a shop capable, i would like to know. Would have to be a featurecam supported shop,

or have machines that i have a G code optimizer for already (i have most of them).

The price on the chassis would definitely go up a bunch. Or people can do what luvdunhill is doing,

which overall is a massive amount of work. On the current chassis, all pieces, there is 12 hours of

machine time. That is just machine time, does not count the added load and unload cycles, or the

cleaning of the goo afterwards. $700 of the chassis was machine time.

But if people still want another board run, that can be done. All fixes have been applied to any

future board run.