There's something strange when using a 10M45s. It cost me about 4 hours work to figure out why my amplifier wasnt working. I used it as a CCS on top of a 2A3. I chose the 100mA current setting resistor according to the diagram (22 ohm). However after searching for about 2-3 hours I found that the CCS only delivered 60mA. So I put in a 10 turn potmeter, and adjusted till it delivered 100mA. Then I measured the potmeter, only to find it at 0 ohms😕😱
I confirmed it after that: with a wire-bridge between G and K it delivers exacly 100mA. So it seems it has already a resistor build in, so it can never deliver more then 100mA. Safety? Ixys made a design change? Did everybody know this already?
Anyway; if you want more then 60mA out of your 10m45s, the diagram isnt accurate anymore, and you can best use a potmeter, and replace it with a fixed resistor representing the same value.
gr. Paul🙂
I confirmed it after that: with a wire-bridge between G and K it delivers exacly 100mA. So it seems it has already a resistor build in, so it can never deliver more then 100mA. Safety? Ixys made a design change? Did everybody know this already?
Anyway; if you want more then 60mA out of your 10m45s, the diagram isnt accurate anymore, and you can best use a potmeter, and replace it with a fixed resistor representing the same value.
gr. Paul🙂
The IXCP10M45 data sheets that I have seen have not been real detailed. I.e. they have not included any information about the performance of the part versus temperature. The max output current of the 10M45 is supposed to be 100 mA so that you are seeing 100 mA with G-K shorted does not chock me, although, the data sheet shows that you should have 20 ohms between G and K. But if the device is warm, that will slide, and Vth increases with increasing temperature so it makes sense that you get less current if the device is warm.
Rather than operating the device at the limit, how about connecting two devices in parallel? Just build them as two independent 50 mA sources and connect both to the 2A3.
~Tom
Rather than operating the device at the limit, how about connecting two devices in parallel? Just build them as two independent 50 mA sources and connect both to the 2A3.
~Tom
Thanx for your reply Tom,
I thought that too, but even with 10 ohms between g-k it was about 75 mA and with 5 ohms about 90 mA. Now with the g-k shorted, it drifts between 99 and 103 mA. 99 when cold, 103 when hot. I'm dropping 180v/ 100mA: thats 18watt dissipation, well below max spec. (it sits on a pentium 4 cooler and even at full 18 watt doesnt get hotter then 40-50 degrees C, see pic)
Dont know about the parallelling of two devices, doesnt it give some unwanted behavior?
Oh and it wasnt a broken chip, because I swapped it, and the new one gives exactly the same behaviour...
I thought that too, but even with 10 ohms between g-k it was about 75 mA and with 5 ohms about 90 mA. Now with the g-k shorted, it drifts between 99 and 103 mA. 99 when cold, 103 when hot. I'm dropping 180v/ 100mA: thats 18watt dissipation, well below max spec. (it sits on a pentium 4 cooler and even at full 18 watt doesnt get hotter then 40-50 degrees C, see pic)
Dont know about the parallelling of two devices, doesnt it give some unwanted behavior?
Oh and it wasnt a broken chip, because I swapped it, and the new one gives exactly the same behaviour...
Attachments
Move to Supertex DN2540 when you can. Has more IDSS and better high frequency. The 10M45S loses high frequency rejection when pushed.
I just bought 25 of the 10M45s...😀 And now you want me to replace them?😉
I thought the 10M45 had better specs? I remember a comparison (tubelab?) between both, and only cascaded did the dn 2540 win.
But isnt it better to build a discrete ccs for these currents?
I was planning that IF the proof of concept amp worked and sounded good. (this still is a prototype. (nr 4) The final one gets a CNC milled case.)
Sorry Salas,
Not very good advice: the Idss of a 2540 is higher, but the Pmax is only 15Watt. It will die very soon in my application. The 10M45S can have 40 watts if cooled accordingly. Vmax of the 2540 is also only 400v, but I need only 360v, so that would just be enough...
18W on a TO-220? I would cascode to halve it anyway. 10M45S has more gate capacitance, not better when single. Also tends to be noisier than the 2540 when in cascode too.
Maybe IXYS doesn't test them very much before they ship the parts. Their data sheet hardly guarantees anything. Most of the specs are typical specs. Graphs are always "typical, not guaranteed". The max output current of 100 mA is not guaranteed as it isn't listed as "output current, min = 100 mA" it's only listed in the marketing blurb at the top of the data sheet.
Because IXYS doesn't actually guarantee anything in the spec sheet, the part you are using does meet the spec.
Get over it. You're not doing anything wrong. Use two in parallel... In the future, if you want predictable performance go with a part that is specified on the key parameters you need for your circuit.
~Tom
according to specs 10m45s in a to220 should be able to handle 40 watts...😱
I wonder if anybody ever used it at that much dissipation...
I wonder if anybody ever used it at that much dissipation...
😀 I wasnt really upset or something...Did the tone of my story sounded like that? That must be some dutch influences, always a problem for me in finding the right "nuances" in english...
I only wanted to inform other people if they are using the part in high current setting, to expect some problems
Don't trust any small part at such levels. Use 1/3 Idss for good Tc. Since you got some, parallel. Now you are using it on saturation max, its not best choice for high end band rejection or Tc.
please pardon me, but why ain't anyone mentioning that these are really gorgeous looking pair of amps!😀
🙂 Thank you! It was meant as a temporary case, made out of cheap MDF, spray painted with "granite spray".
But it turned out too good; even my wife likes them...
Possibly i'll keep 'm just like this, and scrap the CNC aluminium idea... Saves money for the next amp😀
gr. Paul
But it turned out too good; even my wife likes them...
Possibly i'll keep 'm just like this, and scrap the CNC aluminium idea... Saves money for the next amp😀
gr. Paul
Last edited:
Looks very nice indeed, and I like the color scheme a lot too. I just hope that the prolonged exposure to heat will not have a toll on seams and paintwork. Maybe mesh areas for the main valves bases to fix and for circulation can blend nicely too.
I hope that too😉 The underside is completely open, and i'm counting that the large hole in the center for the pentium4 heatsink provides enough airflow...
What he said... But I would calculate the derating factor based on actual case temp.
18 watts is a lot for TO-220 if you want the best reliability. The 40 watt rating is for Tc=25C for which you need Tsink~=16C
My big problem with IXYS is lack of specs. There is a power derating with case temp; for TO-220 MOSFETs it's typically in the range of 0.3 to 1 watt per degree above Tc=25C. At Tc=50C (the pentium cooler might do it but I'd measure) the derating might be conservatively 25 watts, leaving you with 15 watts usable dissipation (for good reliability).
If these are on the anode of a 2A3, there should be no worry at all about the extra capacitance of paralleling 2 devices. Maybe you could use a parallel cascode (4 devices total) and still get to your 100mA.
Michael
PS
If this is a parafeed output stage you could also try tapping the middle of the Rset for the output, which turns the CCS into a complementary output driver (anti-triode). This gives you approximately 1/2 the effective plate resistance compared to the CCS connection, allowing a little more power and bass extension with the same OPT Zpri, or up to 2X power output at a lower Zpri. The parafeed cap value may need to be retuned if you do this. Another way to think about this is that you get to use the power supply voltage drop across the CCS productively, instead of wasting it, as normally done in CCS/parafeed output stages.
Attachments
What is the voltage across the IXYS chip and how is it mounted?
I believe it is better to put run 4 chips in parallel so that each would take approx 25mA that would keep the power dissipation reasonably.
Johnny
I believe it is better to put run 4 chips in parallel so that each would take approx 25mA that would keep the power dissipation reasonably.
Johnny
With language barriers (English isn't my "first" language either, though I find my Danish getting more and more rotten as time goes on), nuances in the English language, and the general deficiencies of email/forum posts it can be challenging at times to figure out the mood or tone of the person who wrote the email/post. That's all good, though. No harm done.
I've had odd issues with the IXYS parts also. In my case, I think I ran into some compliance issues (i.e. the voltage headroom across the part got too low and it stopped regulating). I used it as a load on a triode operating as an input stage in a 2-stage triode-pentode power amp. I found that it caused distortion at even very low power levels when the voltage swing of the amp reduced the drop across the current source. I didn't do much investigation, but replacing the current source with a simple resistor load improved the distortion by almost an order of magnitude and increased the output power by about a factor of four...
I should throw them on a curve tracer one of these days. It's about time the curve tracer at work gets to work at real manly voltages. We normally work in the 3.3 V world and 5 V often gets referred to as "high voltage". 🙂
~Tom
Thanx for the extensive reply!
I just measured it: its 80C😀 ; after playing from 5:30pm till 11:45pm.
It was the maiden voyage for the amp, and it sounds fantastic, much better then all my previous amps. Its the schematic from this post(http://www.diyaudio.com/forums/tubes-valves/158352-dcd-power-amplifier-2a3-d3a-ccs-coupled-new-topology.html#post2042587; just substitute the 2a3 for the 807. The d3a has 20mA current, and 175v Va.
It gave me goosebumps several times, and I had the impression that I heard Ella Fitzgerald standing right before me, and her sax player... very impressive.
But you are right, it is a little much for the poor ccs. I leave it for now, but I think I'm gonna back up just a little. I think 60mA/250v Va will be fine. it leaves me with about 3.5 watt max output power. The CCS has to drop 200v; but only 12 watts dissipation with 200v/60mA. And as a plus I would be able to use a TJ meshplate 2a3, I read great things about them.
The anti-triode setup: very interesting! I'm gonna try it sometimes, and see if I like it. But not in this amp, I would have to rebuild the case completely. There's only room for 2 CSS's on the heatsink, and that room is used up.
Greetz, Paul
Voltage is 180v@ 100mA as stated in the post with the pictures, you can see how its mounted there too.
And youre right, 4 would be better, but I dont have room in this amp...
gr. Paul
- Status
- Not open for further replies.