X100 backengineered here

Buy it from Allied Electronics

These are the approximate prices I obtained from Allied. They are not exact, but I am posting early to enable others to get good prices as well.

IRFP240 -- $2.5
IRFP9240 (these are hard to get from IRF right now) $3.5
IRF610 -- $.5
IRF9610 -- $.65

Method of getting these prices:
1. Go to http://www.irf.com
2. Search for any part number such as IRFP240
3. Check out all the supplier links
4. If one co. has high prices on one part, you can usually haggle the offending part down to the normal level.

In my case, it could have been even cheaper, but I needed the parts now and had to go to those who had all parts in stock.

Now I have 50 of each type of output device and 30 of each type of input device. This should enable totally tight matching.
 
Petter all this is incredible.
Very cool.
I got by the way a demo cd from circuit maker 2000.
It was a 30 day promo and I did something on the computer and now it says date expired.
This has nothing to do with this thread here, but can any body help with this.
There is some file saved somewhere and i can´t make the demo cd work anymore.
Anybody know a crack or something?
 
Change date backwards does not work?????

Install your machine from scratch :-(

Borrow another disk and do it there

Install a dual boot should work

There is a Student demo which should work for this small circuit available for download as I recall without expiry.

Probably many other good fixes.

Good luck
 
A very simple hack that often works (and doesn't involve decompiling the code, with all the attendant hassles & legal questions) is to simply set the date on your computer back a year or whatever.
Just remember to set it back to the present date when you're done.
Jason: Why can I sometimes see other posts...and sometimes not? Right now I can't, for instance.

Grey
 
Actually changing the date doesn't work. CM specifically tells you this when you download the demo.

I'm still in the 30d trial period, so don't know if I will get around this, but I suspect they write to the registry on install, then each day simply decrement some counter (IF today <> yesterday THEN decrement turns_left). This is likely "invisible", ie hidden deep in the registry somewhere where it is not obvious = unlikely to be under a "CircuitMaker" key.

The student version is limited to 30 devices which eliminated all but the really simple circuits.

mark
 
Bummer about the date trick. I no longer have the time to fiddle with reverse-engineering the code. My position these days is just not to mess with crippleware unless I absolutely have to. I don't suppose you could export it to the student version of Spice, could you? (Or some other version of emulator...)

Grey
 
Notes on Thermal Washers

I have researched the field of Thermal Washers for some time now, and made some conclusions which I would like to share with you. I have also ordered the units.

The best thermal performance is from Beryllium Oxide washers. These are toxic when ground up, but should be safe if handled carefully. They appear to only be available in TO3 form factor. I would have got mine from an Aavid distributor such as Future Electronics, but am running TO-247 transistors and so these are out. Just as well perhaps as I understand they are illegal in Europe due to the toxicity of Beryllium dust (when inhaled causes Berylliosis similar to Astbestosis -- funny really that materials so totally opposite in thermal performance have the same characteristics to man).

BeO units are brittle, but since they are thick and non-conductive, they offer improved capacitance effects and should therefore increase high frequency response. They need thermal grease to mount.


The second best alternative (and there really is no need to look furher) is Bergqvist Silpad 2000 which is intended for High performance aerospace/Military/Hi-Rel projects. This is a very high performance film which does not require thermal grease, and so it should be clean to work with (usually people use too muc grease resulting in suboptimal performance anyway). It is also rated to +200C which is useful. Thermal performance really is very very good. The downside is that the film is thin, which will increase the capacitance of the devices used and impede HF response. Having said that, I still went for the thinnest unit since the conductivity still is way off BeO.

I ordered my units from http://www.bergqvistcompany.com and I got part number suffices -122 for my TO-247 units and -58 (-62 was out of stock) for my TO-220 devices. Having a high conductivity film for the low power input devices as well should further assist in keeping junctions at same temperature when hooked up to the same heatsink -- or possibly just to one another wit one piece of film between them.

You may want to order from a distributor as Bergqvist has a $100 per line order minimum. I ended up at $102 for 120 units of -122 and was sampled 15 units of the other type free of charge.
 
Line filter -- best value in town!

I have researched the subject of line filters for some time now, and come to the conclusion that it is very very hard to do better than what you can get from commercial vendors.

There are two types available. Each are oftimized for the type of impedance that each side of where you position the filter is placed. I suspect what we really want from audio is a filter that absorbs what comes from the inside (low impedance to HF) and blocks out from the outside, but for digital polluters we want to block what is generated inside.

The best units cost a lot of money but offer differential mode and common mode attenuation of 60-80dB from 10KHz up. These cost about $60 a piece and are made by http://www.corcom.com -- typically their Q series.

The next best thing is Corcom 3SP1 now out of production similar to 3EP1 (http://www.corcom.com/catalog/filters/EP/Default.htm) with 50dB diff mode and 36dB common mode attenuation from 150KHz ... which is available for $3.25 from http://www.meci.com part number 560-0117. This is a three amp unit which may be a problem in the US, but they are conservatively rated and I should be able to get .75KW out of mine with 240 volt line (divide by 4 for US). The only problem with these is that it is not a power entry module and that it is physically a little large. However if you put one inside your CD player, I guarantee you will hear a difference -- biggest upgrade I ever did, and I used an el-cheapo from an old computer.

So I put my money where my mouth is and bough 10 ...

Mount it, test it and turn it backwards -- I suspect amps will sound best with it backwards and that the system will sound best when digital units are connected normally -- keeping noise from digital polluters way from the line.

[Edited by Petter on 03-08-2001 at 05:02 AM]
 
Petter & Others,

An Aussie amplifier ( http://www.aussieamplifiers.com ) builder ran across my site and we began corresponding re the X_Series. He sent me his version from the patent, which with some minor alteration now looks like this:

http://www.adelaide.net.au/~mefinnis/passlabs/images/x_series_v3.gif

This is a lot closer to Fig.3 from the patent form and follows the normal grounding arrangement. Runs as advertised through Circuitmaker, for those having this the file is at:

http://www.adelaide.net.au/~mefinnis/passlabs/files/xseries_np3.ckt

Obviously for the real thing you would parallel output devices etc.

Comments !! mark

[Edited by mefinnis on 03-09-2001 at 08:20 PM]
 
Anthony's site

Mefinnis,

Excellent stuff!

I am familiar with Anthony's site and used to correspond with him about on the X-series about a year or two ago. Recommended site, nice guy! I hope to start building soon and am getting some help from a local guy who has the practical expertise I am lacking.

Comments:

1. You should seriously consider moving the feedback back to the input stage -- especially if you are expecting current coming back from the amp (I know I am)
2. You should also consider putting some degradation resistors on output stage devices.
3. Interesting voltage source to bias output devices. I suspect though that a variant using physical parameters (Vbe) of BJT's or even fixed bias might be even more effective.
4. Regarding Super Symmetric resistor, I can personally see no reason why we need grounding near it. I would still prefer it at the gates, esp. when considering interfacing to non-floating input devices.
5. You currently have very low voltage gain -- are you expecting a separate pre-input stage or do you have enough headroom from your source?
6. 3A Quiescent current seems a little high -- that's 150W dissipation per device.
 
Petter,

I'll do the easy stuff first.

6&2. 150W/device? No, as stated for the actual amp I would parallel devices. I've put up a pic of what I would consider closer to production, rather than concept.

http://www.adelaide.net.au/~mefinnis/passlabs/images/x_series_v3p.gif

OK, that dealt with .....

1. Feedback position? Please remember I am not an EE, so my theory is not what it could be. This is where NP placed it in the patent, and where he places it for his SE designs, ie from the output? What do you consider the advantages of making FB local to the input rather than global?

4. Re the SS resistor. NP stated in the patent that loss to ground at this point was required to prevent infinite iteration, the "hall of mirrors effect". I'm not sure placing this at the gates has the same effect. Does it?

5. Low gain ..... fair comment.

3. Voltage source = Anthony's (hell man, I'm not that clever!!)

regards, mark
 
I took a look at design posted by mefinnis, and have a couple of comments.

First off, the mosfet output bias is a fairly commmon circuit, same one as used by Pass on the A75.

I'm not sure I understand Petters comment about a BiPolar devices Vbe being a physical parameter. The Vgs of the MOSFET, which this circuit uses as a reference, is also a physical parameter. Certainly there is more device to device variation than Vbe, but I would suspect that in a real world one of the resistors would be a trim pot.

In both this case, and in Petters schematic, this bias circuit needs to be bypassed with a cap.

Of more concern to me, is that this design appears to suffer the same problem as a similar circuit I'm working on. That is the output has a large DC offset. Now the offset is equal on both sides, so there won't be any current flow through the speaker, but it is going to limit the maximum power. I haven't simulated Petters circuit yet, but does it suffer this problem as well?
 
My simulations indicate a higher offset from ground...

Input coupling caps would result in 100% DC feedback, which would lower the offset, but I'd prefer a circuit that naturally zero's itself. I was hoping someone would post a clever solution. Time to pull out a text book or two and don my thinking cap!
 
DC offset and a couple of items

I have a couple of quick points -- will attempt to get back with better answers when I have had time to look at it!

1. DC offset -- You can fix this by twiddling around with potential dividers R36 and R37 on page 1. Of course this is not a problem in that design ... the DC offset you are seeing is essentially the gate drive voltage of the input device.

2. I don't see the need for capacitors across R36/R37 on page 1. These are driven by a constant current source and will thus appear as a voltage source. I guess if one were include it there might not be a significant change, but I don't personally see the need for it. I guess the only way to find out is to try it.



[Edited by Petter on 03-11-2001 at 06:56 PM]
 
Offset Drift

Just a note again on offset issue. From my experience with SOZ topology amp, even though you can adjust the offset to 0 VDC, as the temperature change, the offset will drift anyway unless the MOSFETs have identical thermal characteristics. And the drift value can be so different between the time that the amp is just powered on and half a hour later. This is not much of a problem for the follower, but quite substantial in the first stage.