Finally got time to test my two assembled boards. The first one went well, all the way to TP1-TP2 adjustments via R30. The second one - got stuck at the very first step. Both LED's do light up however voltage across the temporary 10R (fuse replacements) goes up to around 3V and stays there.
Checked around for short circuits - found nothing. Checked random places around C13 and C17 - nothing.
Tried to compare the boards side to side going through them with a DMM, found that resistance across C15 stays around 1.9K when on the good board it goes up to infinity fairly quickly. Retested with DMM at 20K range (instead 2K) and couldn't see the same result (it does go to infinity after a few seconds).
I still removed C15 and verified that it is good. It does appear OK.
Will continue my explorations tomorrow. Meanwhile, if you have an idea of what I should check next - please drop a line. Thanks!
Checked around for short circuits - found nothing. Checked random places around C13 and C17 - nothing.
Tried to compare the boards side to side going through them with a DMM, found that resistance across C15 stays around 1.9K when on the good board it goes up to infinity fairly quickly. Retested with DMM at 20K range (instead 2K) and couldn't see the same result (it does go to infinity after a few seconds).
I still removed C15 and verified that it is good. It does appear OK.
Will continue my explorations tomorrow. Meanwhile, if you have an idea of what I should check next - please drop a line. Thanks!
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I made sure I didn't power it for a prolonged period, and the limiting 10R were always on. I hope they survived.
Anyway, I ultimately discovered I double goofed. I mistakenly swapped all output transistors as well, all PNP and NPN were misplaced. Took them out, put into right places, and now the readings are as expected.
By the way, the hfe's on my Q14 and Q15 are very different. I measured all 4 and here is what I got:
First board (the one I was worried I fried):
MJE15032 shows 90 hfe
MJE15033 shows 185 hfe
Second (good) board:
MJE15032 shows 96 hfe
MJE15033 shows 199 hfe
Is the mismatch a reason for concern? Should I go shopping for a better matching pair of drivers?
Anyway, I ultimately discovered I double goofed. I mistakenly swapped all output transistors as well, all PNP and NPN were misplaced. Took them out, put into right places, and now the readings are as expected.
By the way, the hfe's on my Q14 and Q15 are very different. I measured all 4 and here is what I got:
First board (the one I was worried I fried):
MJE15032 shows 90 hfe
MJE15033 shows 185 hfe
Second (good) board:
MJE15032 shows 96 hfe
MJE15033 shows 199 hfe
Is the mismatch a reason for concern? Should I go shopping for a better matching pair of drivers?
The driver bases use current that flows from VAS to it's current sink.
If the VAS current is taken instead to the upper driver and re-injected by the lower driver, then the VAS current will equal the sink current.
Where the driver hFEs are similar and the base currents are small compared to the sink current, the small discrepancy in the two base current is unimportant.
But where the difference in base current becomes large with respect to the sink current then the effect become significant.
In the extreme, the difference in base current could be similar to the sink current and then the VAS would see no current. It would switch off and the Vbe multiplier would also be incapable of maintaining it's Vbias. That situation cannot be corrected by NFB.
If the VAS current is taken instead to the upper driver and re-injected by the lower driver, then the VAS current will equal the sink current.
Where the driver hFEs are similar and the base currents are small compared to the sink current, the small discrepancy in the two base current is unimportant.
But where the difference in base current becomes large with respect to the sink current then the effect become significant.
In the extreme, the difference in base current could be similar to the sink current and then the VAS would see no current. It would switch off and the Vbe multiplier would also be incapable of maintaining it's Vbias. That situation cannot be corrected by NFB.
I had a test run of my two assembled boards and they performed fine with a 25VDC power supply (to the point where I got sound at the output) however each one blew a pair of output stage transistors when using 66VDC PSU, each time after 20-30 seconds of idle run.
I use NJW0281 / NJW0302 for the power stage and MJE15032 / MJE15033 for drivers. The fact that each board blew in a similar fashion tells me that it is unlikely due to an error with assembly.
One thing that I am suspect of is the driver transistors. I ordered these off of ebay shipped from China. Could poor quality of drivers cause a failure of output stage? Any other ideas?
I use NJW0281 / NJW0302 for the power stage and MJE15032 / MJE15033 for drivers. The fact that each board blew in a similar fashion tells me that it is unlikely due to an error with assembly.
One thing that I am suspect of is the driver transistors. I ordered these off of ebay shipped from China. Could poor quality of drivers cause a failure of output stage? Any other ideas?
Last edited:
Hi reedcat,
You are kidding - right. Buy current semiconductors only from authorised dealers and distributors. Those outputs are no longer available and they are the nicest ones I have seen in a long time. I personally can`t believe you bought off Ebay.
Okay, Ebay parts, throw them out now. Do not use them in anything. Order replacement parts through the proper channels. It would be better if your Ebay parts were compressed with a large hammer before discarding them. If you keep them, they will bite you again and again. It is up to you how much pain you want to live with.
I can`t even imagine why you bought from Ebay. They took longer to get to you, and they are mysteries. Digikey tends to be the most expensive distributor, but I buy there from Canada. Newark tends to have stock in England (why), so Mouser might be a good alternative. At least these parts will be real.
-Chris
You are kidding - right. Buy current semiconductors only from authorised dealers and distributors. Those outputs are no longer available and they are the nicest ones I have seen in a long time. I personally can`t believe you bought off Ebay.
Okay, Ebay parts, throw them out now. Do not use them in anything. Order replacement parts through the proper channels. It would be better if your Ebay parts were compressed with a large hammer before discarding them. If you keep them, they will bite you again and again. It is up to you how much pain you want to live with.
I can`t even imagine why you bought from Ebay. They took longer to get to you, and they are mysteries. Digikey tends to be the most expensive distributor, but I buy there from Canada. Newark tends to have stock in England (why), so Mouser might be a good alternative. At least these parts will be real.
-Chris
Most of the time buy not always. In the past Mouser, DigiKey etc always supplied proof of authenticity when one bought larger quantities (25 for power transistors, 100 for small ones if I remember correctly). Now it's not always the case. The safest is purchasing directly from manufacturers but not all sell small quantities. That's a pity.
Once I took the case to the manufacturer, OnSemi, as I suspected that power transistors I bought were fake. They helped me how to check them properly and when supplier culd not produce the proof of authenticity I sent them back.
cheers,
Once I took the case to the manufacturer, OnSemi, as I suspected that power transistors I bought were fake. They helped me how to check them properly and when supplier culd not produce the proof of authenticity I sent them back.
cheers,
Last edited:
Neither Digikey nor Mouser have MJE15032 in stock. This is why I ordered from eBay. I ordered all other parts from these two sellers.
Why not use MJE15030/31, or34/35? All are stocked by mouser. I was surprised that 15032 is out, and the 26 week delivery from the manufacturer looks terribly long...
I used NJW0xxx outputs, but now my favourite is the Sanken 6145/2223.
Sajti
This sounds very likely you are having fake transistors where ce voltage is not as rated.
I had used fake bd139 as vbe multiplier. AMP blows up in like x weeks up to 3 months. When used as current driver 50v ce, blow up when amp power off.
If you want to test for geniune quality. Load up the transistors at 66v dc and give it half of max rated power while drectly mounted on reasonable cooling with heat sink grease. Should fail within 10secs if fake. Take care if your supply is not short proof, the transistor will blow up if fake and be very scary.
Last edited:
I'm beginning to question my choice of heatsink for Q10-Q12. I have the right size but it's quite thin. I took the piece from an old xbox heatsink and it's thicker at the "bottom" than at the "top". Size is more exactly: 70x26mm and 1,2mm thick at the bottom but only 0.5mm at the top. Unfortunately I've already installed it on one of the pcb's.
Not sure how much heat it is suppose to dissipate. Would appreciate if anyone could give input about this.
Also another "noobish" question. Does the orientation of C1 matter? Using the WIMA MKS4.
Will try to post images of heatsink below. Let me know if more images/description is needed, thanks.
Imgur: The most awesome images on the Internet
Not sure how much heat it is suppose to dissipate. Would appreciate if anyone could give input about this.
Also another "noobish" question. Does the orientation of C1 matter? Using the WIMA MKS4.
Will try to post images of heatsink below. Let me know if more images/description is needed, thanks.
Imgur: The most awesome images on the Internet
Last edited:
Just what to clear this up. The transistors that blow up are output stage NJW0281 / NJW0302 ordered from Digikey. The (possibly) fake ones are drivers MJE15032 / MJE15033. They do not blow up.
Well theres a possiblity that either vbe multiplier transistor is fake, or your idle current becomes much higher when at 66vdc and needs to be dropped by chaning ratio of vbe multiplier.
Driver and output transistors both could also be fake.
Some designs have their idle current slowly increace from start up. They do not instantly bias your transistors at max setting bias from cold start up.
The best way to check is to put your transistors on+-66v =122vdc and load them up.
Driver and output transistors both could also be fake.
Some designs have their idle current slowly increace from start up. They do not instantly bias your transistors at max setting bias from cold start up.
The best way to check is to put your transistors on+-66v =122vdc and load them up.
Last edited:
Hi reedcat,
Such a shame. You could even have a problem with your insulators / grease application or even mounting screw tension. If the drivers are unknown, then they could be fakes and blowing up the output transistors. Remember that the outputs are what take the brunt of the energy.
Ignore any advice to run the transistors at x voltage at y current. To do this, you must look at the S.O.A. curves and ensure you are within safe limits. Next, you have to measure the case temperature (usually through a hole to gain access to the case right under the die mounting). Temperature does come into play here. I've not heard you say that you compared the device markings with the data sheet either (the one on the On Semi web site). You could probably send An Semi a picture for them to rule on it. This is too easy and does not involve destroying any more devices.
If you have a good LCR meter (as in really good, not "Good for the money"), measure a known good one for C-B capacitance and E-B capacitance. Why not check out the thread My Transistors, original or copy?
There is a lot of good information spread throughout that thread and some members have found methods that work for them. It seems exactly right up the alley you would want to read from. SMY14INCH, you should probably read that again. Note that this thread is a "sticky".
-Chris
Such a shame. You could even have a problem with your insulators / grease application or even mounting screw tension. If the drivers are unknown, then they could be fakes and blowing up the output transistors. Remember that the outputs are what take the brunt of the energy.
Ignore any advice to run the transistors at x voltage at y current. To do this, you must look at the S.O.A. curves and ensure you are within safe limits. Next, you have to measure the case temperature (usually through a hole to gain access to the case right under the die mounting). Temperature does come into play here. I've not heard you say that you compared the device markings with the data sheet either (the one on the On Semi web site). You could probably send An Semi a picture for them to rule on it. This is too easy and does not involve destroying any more devices.
If you have a good LCR meter (as in really good, not "Good for the money"), measure a known good one for C-B capacitance and E-B capacitance. Why not check out the thread My Transistors, original or copy?
There is a lot of good information spread throughout that thread and some members have found methods that work for them. It seems exactly right up the alley you would want to read from. SMY14INCH, you should probably read that again. Note that this thread is a "sticky".
-Chris
I heard he said its from digikey the outputs. But still may be fake cause they blew up when drivers stay alive.
Maybe fake but probally not likely
Hi HI-FI PRO,
Yes. However, if the drivers are leaky at those voltages you would lose control over the bias and cook the outputs to death. If it happens quickly, you could still exceed SOA. The outputs could blow without heating the heat sink in that case. What is important is the die temperature and any hot spots that might exist.
I still think that for certain answers, contact ON Semi and show them a picture of the parts.
-Chris
Yes. However, if the drivers are leaky at those voltages you would lose control over the bias and cook the outputs to death. If it happens quickly, you could still exceed SOA. The outputs could blow without heating the heat sink in that case. What is important is the die temperature and any hot spots that might exist.
I still think that for certain answers, contact ON Semi and show them a picture of the parts.
-Chris
