Exicon 10N20 10P20 Transistors - help please.

[seebert]

Trying a new post for the transistors gone in my Muse160 amp:

10N20 and 10P20 transistor sources in the USA anyone? Or alternatives?

Found this source in the UK - same transistors as in the Muse amp?

http://www.profusionplc.com/pro/gex/pcatdtl0?ipartno=ECX10N20R

http://www.profusionplc.com/pro/gex/pcatdtl0?ipartno=ECX10P20R

Thanks, Mike.

(from thread on all my amps ....)

 

[qusp]

afaik the profusion laterals were diffused by semelab, element14/farnell stock quite a few of the semelab stuff under the different brands like alfet. profet i think are too, but regardless you should be able to find something suitable here

your links are broken btw, but i gather they were a 10A 200V part? unless you were really pushing them, the ALF08N20V and its complement might do the trick, but there are higher power double die devices like ALF16N16W

hope that helps

 

[dillmeister]

Rod Elliot recommends these in his P101 project. He shows them as an alternate to 2SK1058 and 2SJ162. I have ordered from Profusion and their service and packaging is great. Their standard shipping is fast but expensive. You could phone them and see if they will go with regular air rather than courier. I hope this helps. ( :

 

[stoc005]

Use FindChips.com Online Electronic Component Distributor Inventory Search. ALF08N/P20 in T03 or TO247 are in stock at Newark in the US. They also carry the ALF16N/P20 in either package. All made by Semelab.

 

[seebert]

Thank you all for your input and great help.

qusp: I've looked at Element 14 but (being a bit slow now) cannot locate the part - it's a confusing site for a non-tech person

dillmeister: Thanks for the OK on Profusion - it's always a bit 'risky' ordering from a supplier overseas - especially one I hadn't previously heard of. I think I'll go with them unless I can figure out Element14 !!

I've altered the links - the forum 'system' automatically put in a an HTML 'URL' link. Sorry.

Thanks again and my best regards and wishes, Mike.

stoc005: Our posts crossed. Many thanks for the info. I haven't looked into it yet - wanted to get a 'thank you' right back to you. Mike.

 

[seebert]

stoc005: Many thanks !!! Yes I see them - that's a weight off my mind. I still don't know why they 'went' in the first place ... all I was doing was checking DC offset with nothing at all connected to the Muse amplifier and when I reconnected the speakers there was bad distortion.

 

[qusp]

Thank you all for your input and great help.

qusp: I've looked at Element 14 but (being a bit slow now) cannot locate the part - it's a confusing site for a non-tech person

mike, i linked you directly to 2 of them at farnell/element14 that could be substituted provided you can handle an: 8A rated device or a 16A rated device vs the 10A rated devices you linked. they are otherwise diffused by the same larger company: semelab but sold under a different name by a different company. the devices themselves are the same lateral fets

 

[seebert]

Sorry qusp, I was still 'rattled' as I hope you can imagine and stupidly didn't realise they were links Since these transistors are expensive (by my standards anyway) can you (or anyone) tell me how to test them. I tried a quick check in-circuit but got confusing readings so will drop them out of circuit this afternoon to check them. I have DMM with diode check facility but don't know what numbers to expect from these transistors. What should I read on a good one from:

PIN 1 Gate to PIN 2 Source or to PIN 3 Drain ... in easy baby talk please, like positive lead to X and negative lead to Y (or any combination as required) should read XYZ figure approx. I know I'm slow but really appreciate all help and I do learn (albeit slowly) from it.

Is it a requirement to change out the whole set (all 8 per side) or just the faulty ones?

By the way, what is the benefit of using lateral fets? I don't know a thing about them compared to 'ordinary' transistors.

Thanks, Mike.

 

[tvrgeek]

Newark had them? Darn. Ordered a set from England yesterday. Although Newark is pretty pricey, and postage was quite reasonable.

Anyway, Mike. You can't really test a MOSFET like a simple bipolar with a multimeter. Google how to build a test fixture. Be VERY careful as they are ESD, as in static, sensitive. What makes a lateral MOSFET better than any other device seems to be up for discussion and I am not qualified to answer. Mr. Curl seems to hold different views from Mr. Self, than Mr. Pass, than...... All very bright gentlemen. As I am repairing a DH 120, I am using Lateral MOSFETS. It is what it was designed for. Will I when I try and build my own? Not a clue. I suspect the execution is far more important than the part until you get to the level of the folks fore mentioned. I like my MOSFET Hafler and B&K, but also my BJT Rotels.

 

[zinsula]

These are lateral MOSFETS. No need to replace all if only one is broken. They have such a tempco that they will "equalize" themselves regarding the quiescent current even if you do not match them.
I bought from profusion recently, no problem for them shipping to Switzerland. And in my case way cheaper than Farnell.

In the ALFET, Magnatec, Exicon lateral mosfets, it's always the same chip.
You can trust him: http://www.diyaudio.com/forums/anal...rch-preamplifier-part-ii-275.html#post2634030
last sentence

 

[AndrewT]

..........lateral MOSFETS.................They have such a tempco that they will "equalize" themselves regarding the quiescent current even if you do not match them...........

not the case.

Even with very accurate selection of Vgs (at the actual operational quiescent current) and very accurate selection of Rs, you will find on power up that the drain current has tiny non equalities between devices.

A small Vgs error in the initial matching results in a disproportionate Id mismatch in the operating amplifier.

A big Vgs error in the initial selection will result on that device dissipating a much different Pq from it's brothers that it will run at a vastly different temperature (Tc & Tj) from it's brothers.

 

[zinsula]

[...]A big Vgs error in the initial selection will result on that device dissipating a much different Pq from it's brothers that it will run at a vastly different temperature (Tc & Tj) from it's brothers.

With vertical MOSFETS yes, absolutely. With laterals like these Semelab parts no, my experience differs very much.

1. These parts need to be biased with at least 150mA per device anyway to get them in a more linear transfer curve region. Coincidently, this is just above the zero tempco current. If one part runs away, it needs more voltage, so it is self stabilising
2. Transconductance of these parts is so low compared to verticals, you can model it as a source resistor of way more than 1 Ohm (looking at the graph, around 150mA, more like 5 Ohms!!) at these low currents. Don't tell me that the bias currents will differ wildly.

Get some parts, measure Vgs (you'll also find out that they match better than verticals), mount the two most different parts on a heatsink, bias one with 150mA an and report back what the bias current of the other is.

 

[seebert]

Would appreciate your comments on an email I just received from a company whom I asked to repair the Muse 160 amp:

The Big EXICON 10N20 and 10P20 MOSFET transistors are no longer made and there are very many fakes available on the net.
One or more of these MOSFET transistors is defective, which is causing an extra load on the small driver transistor (causing it to overheat). As much as we would love to repair this unit for you, we would not be able to due to the obsolete MOSFET transistors used.
There are no equivalent matching units that we have come across and some folks on the web sell these for $40+ per pair.
In your case, the complete set (8 units) would need to be replaced to maintain a balance.

I then sent them info from this post and get this email back:

HI Mike

Thanks for the followup. We have been in touch with Profusion PLC and they informed us that the MOSFET was no longer made. The alternatives you suggested will work but are not the exact replacements.
Therefore, if you want to go ahead with the 8A modules, then you should replace these on both channels to balance the audio.

Regards,

So if Profusion is out of the original Exicon's is the rest of the info okay to go on - and would you go with the lower power ones: SEMELAB - ALF08N16V - MOSFET, AUDIO, N, 160V, 8A, TO-247 or the higher power ones : SEMELAB - ALF16N20W - MOSFET,AUDIO,N CH,200V,16A,TO-264

Seems that I am going to have to try a repair myself since nobody wants the job

Thanks and best regards to all, Mike.

 

[washburn_it]

First days of January I wrote to Profusion since I needed to buy 10 ECX10N20/10P20 (I'm building the P101 of Rod Elliott) but they were out of stock.
This is the reply that I had from Profusion:

"We currently have a serious supply problem with the ECX10N20R/ECX10P20R. Our best estimate on new stock is early April."

I found them over the net from a person that bought 8 of them a year ago so we went to a deal and he sent me those transistors.
Hopefully in April they could have them in the stock again....

Regards,

Roberto

 

[tvrgeek]

The Semilab also have the gate zieners the big Exicon's don't.
The Exicons are listed in their current line catalog and on the WEB. I gust got 4 of the smaller ones.

 

[seebert]

Many thanks washburn_it (Roberto) - that's good to know - it sounded as though they were gone forever.

tvrgeek: The Exicons are listed in their current line catalog and on the WEB. I gust got 4 of the smaller ones.

Great news !! Could you please post me a link (again) to the Semilab and Exicons that are listed. I'm so sorry that I am now so slow in my thinking - I had another fall last week and am pretty much dumbed down with medication.

True thanks and my best to you all, Mike.

 

[SAINTGERMAINMUSIQUE]

you can replace the Mosfet EXICON 10N20 by a 2SK176 or a BUZ901. F

 

[SAINTGERMAINMUSIQUE]

exicon 10N20

you can replace the EXICON 10N20 (EC 10N20 for datasheet) by a 2SK176 or a BUZ901;
For the EXICON 10P20, you can replace by 2SJ56 or a BUZ906;
These osfet transistor are present in Q6000 Chevin amplifier

 

[ilimzn]

With vertical MOSFETS yes, absolutely. With laterals like these Semelab parts no, my experience differs very much.

1. These parts need to be biased with at least 150mA per device anyway to get them in a more linear transfer curve region. Coincidently, this is just above the zero tempco current. If one part runs away, it needs more voltage, so it is self stabilising
2. Transconductance of these parts is so low compared to verticals, you can model it as a source resistor of way more than 1 Ohm (looking at the graph, around 150mA, more like 5 Ohms!!) at these low currents. Don't tell me that the bias currents will differ wildly.

Get some parts, measure Vgs (you'll also find out that they match better than verticals), mount the two most different parts on a heatsink, bias one with 150mA an and report back what the bias current of the other is.

Actually, AndrewT is right, although the consequences are not as dire. I have recently used and selected hundreds of the Exicon parts, also original Hitachi 2SJ56/2SK176 - there is something to be gained by selecting them.
Namely, it''s not that much of a problem at low currents where currents will equalise but you do get a 'jagged' transconductance if your treshold voltages are not fairly evenly matched. In particular, Pch parts can have a very wide spread, and you can get differences up to 1:2 in currents - even above 150mA per device. This makes for a rather uneven transfer curve tat low currents and in some cases it can be VERY obvious (a glaring example would be a hybrid amp using the MOFETs as a follower power stage). THe real difference tends to happen at clipping - the lowest treshold, highest gm device will be the most stressed and the differences again can be quite high - and again on Pch parts mostly. In my particular experience if you are not careful, a 1:2 current mismatch on the P side is possible though not typical at all. In essence, this boils down to the following: with no selection, you can typically expect that 3 random pairs in parallel will do the work of 2 'ideally selected' ones in parallel. Using parts from the same batch considerably improves the situation, to about 4 random pairs doing the work of 3 ideally selected. Selecting for similar treshold and Vgs at a higher current (say 1A) will get you to that point as well, as parameters track to a large extent if matched at two points. If you have a curve tracer or similar device, it's possible to select parts quite well and expect almost ideal scaling of performance as you parallel the parts, but the wide spread of Pch part parameters can be a problem. In general it's possible to find almost a perfect pair out of 10 Nch parts, whereas expect to need about 25 parts to find a perfect pair of Pch parts. Modern versions (Semelab dies and new Renesas dies) are better in this regard, the data above was gleaned by sorting through a large number of original Hitachi TO3 parts (about 20kg worth )...