Stax amps parts specs

[Earspeakers]

I'd like to call into question some do's and don't I've heard about building Stax amps. I'm an engineer who works on scientific instrumentation and have heard a few things I don't understand based on my experience, so thought it's worth a discussion here. Please correct me if my thoughts on these issues is incorrect, from an engineering perspective. 

 

High Voltage resistors are necessary

 

In general HV boutique resistors are really only needed when there’s a large differential voltage. Let's take a high quality but not HV resistor, the KOA Speer which range from 250V-1kV working voltage

 

http://www.koaspeer.com/catimages/Products/MF-MFS-RK/MF-MFS-RK.pdf

 

    Take the average as 350V. Looking at the Megatron schematic I’d have to calculate the voltages involved based on the currents running through these resistors (good old Ohms law), but the ones to look out for are like R12/R24 which go to ground. Those are coming off the cathodes and are biased from that and the 1M coming off the grid. Oh R8 might see something too. The rest of them are only going to see whatever voltage they drop which shouldn't be much. But the whole amp is at best 450V (less in reality), so what resistors in the amp are going to have to bear more than 350V working voltage? I'm not seeing a lot jumping out at me, but I wouldn't mind running a simulation of this circuit to see. 

 

    The other concern might be insulation breakdown to a nearby ground plane. They list the insulation resistance it as not less than 10G Ohm@100V, and the important parameter we are looking for is Dielectric Withstanding Voltage ("The rated voltage that can be applied to a designated point between the resistive element and the outer coating, or the resistive element and the mounting surface, without causing dielectric breakdown.”), which is listed as no breakdown. 

 

Thick traces

 

We had this discussion on another thread but I thought to include it here. 1Oz traces is standard in the business. 1/2 Oz is used too in the high frequency stuff I do. Now 450V isn't really considered high voltage, that's more when you get above 1kV, say 5kV which isn't uncommon in the days of picture tubes. I've worked at national labs in High Energy physics where we used many tens of kV for Silly/Gelly detectors and such. At any rate for those you have to start worrying about corona, so putting more height on your trace is undesirable. The important consideration here is margins and trace width. 

 

Otherwise I've heard it's easier for DIY work, and I'll start by saying I haven't compared a 1Oz to a 2Oz board back to back on this. But again I don't see why more is better. It's mouse nuts compared to the solder our ham fists are going to flow into the joint. 

 

The only reason I know of for thick traces is if your running a lot of current which isn't the case here. 

 

 

Ceramic Insulators

 

Let's look at the data sheet for a typical TO-200 thermal pad commonly available. The Bond-Ply® LMS-HD is listed with a breakdown voltage of 5k VAC. Now the liquid version right below that is much less, 250V/mil, and it depends on how it's applied. The same goes for the silicon pad too, over tightening will reduce these margins, but as far as I can tell they're within spec for the voltages we're working with. 

 

On the ceramic the specs aren't so clear to me, "21.7 x 103 volts/mm". Not sure how to interpret that, if anybody knows I'm interested. But surely the answer is a very high voltage, in the kV range I'd expect. 

 

Conclusion

 

While I'm an experienced engineer I'm new to building Stax amps so am open to missing some points here. In general I'd say it's not true that over-engineering doesn't hurt, many times it actually is counter productive. But it is generally true that having a large margin/spec is good engineering. In this case I think that over speccing these components isn't a bad thing, though possibly it has marginal benefit. 

 

 

[Pars]

I've never built a Stax or electrostat amp, but I'll take a swing at a couple of things you bring up:

 

HV resistors: I think some people had problems with leakage in particularly the battery section of the T2. They found that the xicon 273 series, raised off the board alleviated that. From what I recall, PRPs exhibited this problem as well as the seemingly beloved Dales. Raising them above the board would seem to indicate symptoms of breakdown to the ever present ground plane below them. They are cheap (the xicons), so if that is what works, it's not like a Black Gate mantra...

 

Thick traces: not sure where the now std. 2 oz copper came from. Back when I was more involved in board group buys, usually from Imagineering, 1 oz. was standard.

 

Insulating pads and fasteners (PEEK screws) seems to have come about from a what works perspective.

 

Realize that people like Kevin Gilmore, Craig Sawyers, etc. that are instrumental in these threads are also engineers.

[spritzer]

The non engineer will chime in then.    We don't specify HV resistors except where they are actually needed.  I for one don't count the Vishay/Dales as a high voltage type even though they are good for 500V.  The ballast resistor should be a 1000V unit and the output resistors should be 500V or more. 

 

The thicker traces aren't needed per se but they also don't hurt.  Kevin and I use larger traces than is usual so making those thicker will only help with the thermal performance of the boards i.e. when soldering and as is the case with DIY stuff... desoldering. 

 

Ceramic insulators are simply the end result of lessons learned the hard way.  The thinner insulators can and will be damaged.  For instance when using the IXYS ccs on the KGSSHV the tab of the transistor will see a lot of voltage so why not go into overkill mode. 

 

I wouldn't really call this over engineering in any way, this is the sensible sort not driven by any cost factor.  Overkill is to use a 400VA transformer where a 100VA would suffice and drive the pass transistors in the psu into oblivion. 

[kevin gilmore]

For filament lines, best to have 2 or 3 oz copper, so either you make the trace a bunch

wider, or you get thicker copper. Amperes of filament current...

 

the high voltage eventually jumps around the insulator, so best to use the ceramic

insulator with the special feedthru that goes along with it. then and only then can you use

the metal screws.

[Earspeakers]

Realize that people like Kevin Gilmore, Craig Sawyers, etc. that are instrumental in these threads are also engineers.

 

I'm not saying otherwise and am definitely not pointing fingers, also not doing an appeal to authority here. Just wanting to get clarity on some points, and highlight these safety points if they're important. Good information otherwise, thanks. 

 

For filament lines, best to have 2 or 3 oz copper, so either you make the trace a bunch

wider, or you get thicker copper. Amperes of filament current...

 

the high voltage eventually jumps around the insulator, so best to use the ceramic

insulator with the special feedthru that goes along with it. then and only then can you use

the metal screws.

 

Both points make perfect sense. Specs are one thing, practice is the final arbiter. 

 

However what is the "special feedthru that goes along with it"? In my lab stock at work we have plastic bushings for TO-220 screws, do you mean those or something different? 

 

I wouldn't really call this over engineering in any way, this is the sensible sort not driven by any cost factor.  Overkill is to use a 400VA transformer where a 100VA would suffice and drive the pass transistors in the psu into oblivion. 

 

Agree, good points. 

 

Along these lines, these are the safety points I've picked up so far, any others to put on the list?

[spritzer]

You need a special bushing for the ceramic insulators.  Can't remember the part name though...

[Earspeakers]

You need a special bushing for the ceramic insulators.  Can't remember the part name though...

 

Hm, yeah searching isn't bringing up anything obvious. However the solution my buddy uses is to use PEEK screws. With that I assume any old plastic bushing would be fine. 

[spritzer]

On none at all and just a plastic washer. 

[Earspeakers]

On none at all and just a plastic washer. 

 

Yeah exactly, the bushing just makes everything line up nicely.

[mwl168]

Apologize for the ignorant question;

 

Does one need to use peek nuts with peek screws in this application? The only reason I ask is the cost - those peek nuts are expensive.

[spritzer]

I use M3 standoffs instead of nuts.  Less chance of the thread stripping.

[mwl168]

Good point about the thread stripping. I assume the standoffs do not need to be peek material?

[kevin gilmore]

you want the pps washers

http://www.aavid.com/product-group/accessories/washers

 

they go most of the way into the ceramic insulator and prevent

arc's thru the center of the insulator

 

pretty sure I use the 7721-300sg

will look later

Edited March 1, 2015 by kevin gilmore

[Craig Sawyers]

Ineed Kevin.  

 

The main issue is not the withstand voltage of the insulating washer so much as the fact there is a hole in it through which the bolt goes,  If you draw out a cross section, the distance between the transistor tab and the heatsink is just the thickness of the washer, in air.  So maybe 0.1-0.2mm - which definitely breaks down, regardless of the length of the bush.

 

The only way around this is to use a thick insulating washer - and that is the only reason for using the ceramic ones - to increase the physical distance.  Having done that the next weakest point is around the insulating bush to the mounting bolt. Hence the long AAVID ones that go most of the way through the ceramic insulator.

 

Problem with PEEK screws is if you look at the thermal resistance between a TO220 transistor and a heatsink, a certain minimum tightning torque for the screw is necessary.  You can't get even close to that with PEEK because the threads strip.  Been there, done that, spent the money and have a box of spare PEEK M3 screws and nuts.  However, with the ceramic insulators and long AAVID bushes, you can use stainless steel screws with absolutely no risk of arcing, and get the right torque.  Because I am a bit anal about this I have a TorqueLeader torque screwdriver for precisely this application.  My T2 a la Gilmore design has been working with that arrangement for three+ years without missing a beat.

 

Craig

[spritzer]

I think the only reason why my T2 arced through the bushing into the screw is because of the hole on the 3675's  isn't quite large enough so I had to force them.  Might have damaged one in the process. 

[mwl168]

you want the pps washers

http://www.aavid.com/product-group/accessories/washers

 

they go most of the way into the ceramic insulator and prevent

arc's thru the center of the insulator

 

pretty sure I use the 7721-300sg

will look later

 

Thanks Craig and all. This is very good to know.

 

Kevin:

 

Do you mean 7721-3PPSG as this one ?  http://www.mouser.com/ProductDetail/Aavid-Thermalloy/7721-3PPSG/?qs=sGAEpiMZZMu6TJb8E8Cjry%252brZ0Ynb0Cb Mouser part number  532-7721-3PPS

[Laowei]

Also if there is interference, verify with calipers that Mouser sent the correct part. I ordered 7721-3PPSG and received a bag marked the same, but the actual parts were 7721-1PPSG with a slightly larger diameter than will fit a TO-220 case. I wrongly drilled out the heatsink to fit the bushing and ended up arcng to ground after a few hours run time. After I rebuild with new ordered parts and checking actual dimensions, everything fit together without modification and now all is working well.

Edited March 1, 2015 by Laowei

[Pars]

Not to mention that the sand in a lot of these amps is getting very difficult and expensive to find. No sense in saving a couple of $ in hardware when it could cost you big $$$ if something catastrophic occurs, not to mention the probably huge amount of time finding the replacements and verifying that they are genuine.

[Craig Sawyers]

I think the only reason why my T2 arced through the bushing into the screw is because of the hole on the 3675's  isn't quite large enough so I had to force them.  Might have damaged one in the process. 

Yeah.  It is interesting to look at the tolerance stack up.  If the transistors have the correct nominal diameter hole, the AAVID bush will go through.  But because punch tools wear, transistors start off with holes at the top of the tolerance range and the bush fits nicely.  When the punch reaches the end of life and has worn smaller, the bush is tight.  The transistors in my T2 spanned this range.

 

Typical spec seems to be 3.6mm +0.1/-0.08mm, so a range from 3.52 to 3.7.  The AAVID bush has a nominal diameter of 3.56mm, so should only be tight with the TO220 hole towards the bottom of its tolerance range.  The bush has an (unspecified) tolerance itself of course

 

In hindsight I should have bought a taper reamer and taken the holes out a tad so the bush was always a nice fit.  But since my T2 continues to work it must be all OK anyway.

[Earspeakers]

Thanks Craig and all. This is very good to know.

 

Kevin:

 

Do you mean 7721-3PPSG as this one ?  http://www.mouser.com/ProductDetail/Aavid-Thermalloy/7721-3PPSG/?qs=sGAEpiMZZMu6TJb8E8Cjry%252brZ0Ynb0Cb Mouser part number  532-7721-3PPS

 

Can Kevin or somebody verify this part number? Thanks  

[Earspeakers]

Can Kevin or somebody verify this part number? Thanks

Realized I could figure this out as I have a caliper and an example bushing.

 

TO-200 tab is about 1.33mm

Ceramic oxide insulator is 1.78mm

 

3.11mm to get you to the heat sink

The ceramic insulator and TO-220 hole diameter is about 3.68mm, but the heatsink diameter is less than that, so the bushing will stop at the heatsink. 

 

So length of bushing should be about 3.11mm, that part is 3.18 which is about right, so yes that part should work. 

 

Oddly the bushing is not meant to go through the heatsink hole, so the screw (if metal) will be at heatsink potential. 

 

[GeorgeP]

why would you need the bushing to go past the ceramic insulator?

[spritzer]

Indeed, doesn't make any sense. 

[Craig Sawyers]

My bag of ceramic insulators, left over from the T2 measure 2.03mm.  Add TO220 tab of 1.33mm and get 3.36mm < 3.18mm.  So all is fine and hunky dory - the bush stops 0.18mm before the heatsink.

[Earspeakers]

why would you need the bushing to go past the ceramic insulator?

 

 

Indeed, doesn't make any sense. 

 

My memory from the last time I used these (some years ago) was that the bushing additionally went through the heatsink. Combined with a washer on the other side, the screw was insulated from both the silicon and the heatsink. Extra degree of protection possibly. I could be misremembering but I believe I did this at least once.