OK thank you for Amp-guts pics. Yes it looks like 8 to 6 ratio. Odd IMO
Next you say no load on output the amp does not fail ? Is this correct ?
If so, then no signal and no load please post what is the DC output voltage present on the speaker terminals. A simple DC offset check, it should be very low and if the gate turn on is correct then very low DC should be present.
Next you say no load on output the amp does not fail ? Is this correct ?
If so, then no signal and no load please post what is the DC output voltage present on the speaker terminals. A simple DC offset check, it should be very low and if the gate turn on is correct then very low DC should be present.
Exactly, with no load, the amp works.
Work with small signal, medium, high..and clipping.
When there's clipping, 2 LEDs smd turn on
There's no DC offset.
I don't remember exact value, but with o-scope, there isn't difference between ac-coupled and dc-coupled (= no visibile offset).
Tomorrow i'll measure with multimeter the exact value.
However, amp has DC protection.
Work with small signal, medium, high..and clipping.
When there's clipping, 2 LEDs smd turn on
There's no DC offset.
I don't remember exact value, but with o-scope, there isn't difference between ac-coupled and dc-coupled (= no visibile offset).
Tomorrow i'll measure with multimeter the exact value.
However, amp has DC protection.
djQUAN could you please verify for me that the output stage is a mismatched number of devices like pergo describes? I am very puzzled why this is. In my mind I can see no reason why the 8 P devices to 6 N devices ratio. This does not look right to me and my memory of all other amps I have worked on... Many thanks......Ciao
yep. they're mismatched in mine too. I originally thought that this was an all N channel amp but after looking closely, they indeed have 8 P channel and 6 N channel devices. all devices are connected to the output via a load sharing resistor so all of them are output transistors.
Sorry to hear you got the flu, I hope it subsides very quickly. I can wait till your better just repost in this thread or PM me, and we can pick up on this after your feeling better.....
The testing with musical/signal source is blinding weather or not the supply is being pulsed at failure and weather or not there was a slow rise in bias idle current draw just before the failure.
Perhaps we should just run the amp into a load without any signal drive for a while and monitor the idle current draw. If this does not fail the amp then we are back to the drive signal possibly causing the breakdown in the channel, and possibly causing the N side to fail like it is. < just thinking out loud here, and what is the 12 volt current draw at Zero bias on the output stage ?
I have seen bias circuitry do some very strange things in my lifetime, and this is sounding more and more like a bias/drive related failure. But this is a experienced guess at this point...
The testing with musical/signal source is blinding weather or not the supply is being pulsed at failure and weather or not there was a slow rise in bias idle current draw just before the failure.
Perhaps we should just run the amp into a load without any signal drive for a while and monitor the idle current draw. If this does not fail the amp then we are back to the drive signal possibly causing the breakdown in the channel, and possibly causing the N side to fail like it is. < just thinking out loud here, and what is the 12 volt current draw at Zero bias on the output stage ?
I have seen bias circuitry do some very strange things in my lifetime, and this is sounding more and more like a bias/drive related failure. But this is a experienced guess at this point...
thank you for the support
today i feel better
I'll post some pics in 3 hours
Before every failure, i see, in my Nissei 60A power supply (80A max) that the current at load (50hz signal, 10A drawing @4ohm @50W) remains costant up to few fraction of second before protection goes on (=failure).
In this fraction before protection, the current rises, but, with large scale amperometer (1A/div), the current rises +1A..so 11A before failure.
Other thing i've seen, that, at 10A costant drawing, the current oscillate a little, but very very little. Note, the amperometer is analog..not digital.
I have a PS with digital amperometer, but it's only 5A max
Without load, the idle current is 1.1A with minimum bias (like now), 1.4A with original bias by Audison, and +infinite Ampere bias at max
I set minimum bias for safe and for tests..but, with any bias current (1.1, 1.4, 2, 2.5, 3A) it fails like always fails.
Without load, if i go on clipping (2 smd LEDs indicate clipping), the idle current rises from 1.1A, to 2.5A
For my experience, at clipping, the current rises a bit...not 1.1A --> 2.5A
The bias circuit hasn't any thermistor (there is a thermistor, but for temp protection in the PWM section), and no Vbe multiplier on chassis.
There's a Vbe multiplier SMD, but it's not mount in the chassis or "sandwitch" with output stage.
the only things on chassis are: output mosfet, PS mosfet, rectifiers, LM7812/7912, and a thermistor for overtemp protection.
i check resistance from chassis, to thermal pad...always infinite
i check output Zobel...ok
I try to find IRF640 in my country. A shop have small quantities from ST.
So, if i'll but that, i have IRF640 from ST + IRF9640PbF by Vishay..
If i buy Vishay N's, i've to wait 10 days...or buy 50 pieces if i want them in 2 days
what do you think?
today i feel better
I'll post some pics in 3 hours
Before every failure, i see, in my Nissei 60A power supply (80A max) that the current at load (50hz signal, 10A drawing @4ohm @50W) remains costant up to few fraction of second before protection goes on (=failure).
In this fraction before protection, the current rises, but, with large scale amperometer (1A/div), the current rises +1A..so 11A before failure.
Other thing i've seen, that, at 10A costant drawing, the current oscillate a little, but very very little. Note, the amperometer is analog..not digital.
I have a PS with digital amperometer, but it's only 5A max
Without load, the idle current is 1.1A with minimum bias (like now), 1.4A with original bias by Audison, and +infinite Ampere bias at max
I set minimum bias for safe and for tests..but, with any bias current (1.1, 1.4, 2, 2.5, 3A) it fails like always fails.
Without load, if i go on clipping (2 smd LEDs indicate clipping), the idle current rises from 1.1A, to 2.5A
For my experience, at clipping, the current rises a bit...not 1.1A --> 2.5A
The bias circuit hasn't any thermistor (there is a thermistor, but for temp protection in the PWM section), and no Vbe multiplier on chassis.
There's a Vbe multiplier SMD, but it's not mount in the chassis or "sandwitch" with output stage.
the only things on chassis are: output mosfet, PS mosfet, rectifiers, LM7812/7912, and a thermistor for overtemp protection.
i check resistance from chassis, to thermal pad...always infinite
i check output Zobel...ok
I try to find IRF640 in my country. A shop have small quantities from ST.
So, if i'll but that, i have IRF640 from ST + IRF9640PbF by Vishay..
If i buy Vishay N's, i've to wait 10 days
...or buy 50 pieces if i want them in 2 dayswhat do you think?
Last edited:
some pics
analog section - crossover - phase - subsonic - etc:
http://img683.imageshack.us/img683/5071/analogsection.jpg
driver section:
http://img535.imageshack.us/img535/4537/drivers1.jpg
http://img146.imageshack.us/img146/2340/drivers1mod.jpg
http://img299.imageshack.us/img299/2632/drivers2.jpg
http://img705.imageshack.us/img705/3382/drivers3.jpg
http://img84.imageshack.us/img84/1607/drivers4.jpg
N channel (now IRF640NPbf by IR)
http://img710.imageshack.us/img710/8376/nchannelzoom.jpg
http://img208.imageshack.us/img208/7061/nchannelzoommod.jpg
http://img208.imageshack.us/img208/2484/nchannel.jpg
http://img291.imageshack.us/img291/6889/nchannel2.jpg
P channel (now IRF9640Pbf by Vishay)
http://img413.imageshack.us/img413/9503/pchannelzoom.jpg
http://img267.imageshack.us/img267/8798/pchannelzoommod.jpg
http://img101.imageshack.us/img101/4600/pchannelzoom2.jpg
http://img263.imageshack.us/img263/2783/pchannel.jpg
power supply view
http://img249.imageshack.us/img249/7402/powersupply.jpg
analog section - crossover - phase - subsonic - etc:
http://img683.imageshack.us/img683/5071/analogsection.jpg
driver section:
http://img535.imageshack.us/img535/4537/drivers1.jpg
http://img146.imageshack.us/img146/2340/drivers1mod.jpg
http://img299.imageshack.us/img299/2632/drivers2.jpg
http://img705.imageshack.us/img705/3382/drivers3.jpg
http://img84.imageshack.us/img84/1607/drivers4.jpg
N channel (now IRF640NPbf by IR)
http://img710.imageshack.us/img710/8376/nchannelzoom.jpg
http://img208.imageshack.us/img208/7061/nchannelzoommod.jpg
http://img208.imageshack.us/img208/2484/nchannel.jpg
http://img291.imageshack.us/img291/6889/nchannel2.jpg
P channel (now IRF9640Pbf by Vishay)
http://img413.imageshack.us/img413/9503/pchannelzoom.jpg
http://img267.imageshack.us/img267/8798/pchannelzoommod.jpg
http://img101.imageshack.us/img101/4600/pchannelzoom2.jpg
http://img263.imageshack.us/img263/2783/pchannel.jpg
power supply view
http://img249.imageshack.us/img249/7402/powersupply.jpg
Have you done any high/low temp < thermal stability >testing of the amp while it is sinked and no load and no signal ?
Can you test no load and no signal with a small hot air heat source on circuitry to see if a little extra heat will induce failure, and or odd behavior ?
At the point we are at now I like to test with freeze spray and a reflow station hot air wand set to low heat to find any thermally sensitive devices.
We are looking for a good reason what would cause all the fets in one group to enhance the way they are to failure of the weakest fet.
Please measure with a voltmeter on DC range the DC offset at the speaker terminals. If possible leave the meter hooked up permanently so it can be read constantly. If it does not change then this tells a lot to me. But I would expect to see some reading even if just a few milli-volts DC.
the pics are fine thank you. By the remarks you must know this amp like the back of your hand now
Can you test no load and no signal with a small hot air heat source on circuitry to see if a little extra heat will induce failure, and or odd behavior ?
At the point we are at now I like to test with freeze spray and a reflow station hot air wand set to low heat to find any thermally sensitive devices.
We are looking for a good reason what would cause all the fets in one group to enhance the way they are to failure of the weakest fet.
Please measure with a voltmeter on DC range the DC offset at the speaker terminals. If possible leave the meter hooked up permanently so it can be read constantly. If it does not change then this tells a lot to me. But I would expect to see some reading even if just a few milli-volts DC.
the pics are fine thank you. By the remarks you must know this amp like the back of your hand now
you can say it loud!!
tomorrow i'll test continuosly the DC offset at idle, with no/yes signal and no/yes 4ohm resistive load.
So i can draw a sort of a graph for DC offset
Then, if remains good FET (1 failure every time..and i've still 5 good replaces), i'll try to heat the drivers' zone with hot air (thanks to my desolder hot air station).
In this week, i'll buy new original irf640 by ST or Vishay from original-to-original replacement
thank you!
tomorrow i'll test continuosly the DC offset at idle, with no/yes signal and no/yes 4ohm resistive load.
So i can draw a sort of a graph for DC offset
Then, if remains good FET (1 failure every time..and i've still 5 good replaces), i'll try to heat the drivers' zone with hot air (thanks to my desolder hot air station
).In this week, i'll buy new original irf640 by ST or Vishay from original-to-original replacement
thank you!
i'm making some tests.
All this test are without chassis.
Only hot part are the drivers..very very hot, i can't touch it after a minute.
Output stage is cold.
At minimum bias, set by adj pot, i've:
- Idle current: 1.3A stable
- DC Offset: 12mV without load, 11.8mV with load
- Gate voltage: 2.36V for N-mosfet, -2,36V for P-channel.
If i increase bias:
- Idle current: increase..then goes to +infinite at max bias (5A autolimited by power supply)
- DC Offset: still 12mV, no changes
- Gate Voltage: increase proportionately to bias (obviously)
Then, i take an hairdrier and heat for 2 minutes the drivers zone.
The DC offset: decrease from 12mV to 7mV
The idle current increase 20mA, so 1.32A
All this test are without chassis.
Only hot part are the drivers..very very hot, i can't touch it after a minute.
Output stage is cold.
At minimum bias, set by adj pot, i've:
- Idle current: 1.3A stable
- DC Offset: 12mV without load, 11.8mV with load
- Gate voltage: 2.36V for N-mosfet, -2,36V for P-channel.
If i increase bias:
- Idle current: increase..then goes to +infinite at max bias (5A autolimited by power supply)
- DC Offset: still 12mV, no changes
- Gate Voltage: increase proportionately to bias (obviously)
Then, i take an hairdrier and heat for 2 minutes the drivers zone.
The DC offset: decrease from 12mV to 7mV
The idle current increase 20mA, so 1.32A
now i remember that the first time i've repaired this Audison, i use IRF640 by ST from Taiwan.
http://www.datasheetcatalog.org/datasheet/stmicroelectronics/3007.pdf
When it fails, after playing 10 minutes on subwoofer at high volume @1ohm (Re audio), i think Taiwan = fake
i think it was fake because the package was different to220..but i never check datasheet for confirm...i just buy new IRF640N from IR...
Today, searching for my mosfet collection, i find those IRF640. They are only 4 (10 bought - 6 used).
Looking the datasheet specs, i was wrong...i don't think they're fake (i hope).
The string over mosfet said:
IRF640
CC0FG W
MRC 436
(+ ST logo)
Original P-channel is IRF9640 from IR, now IRF9640PbF from Vishay.
Today i can try those 4 IRF640 by ST @4ohm @low power, just for try.
I think 4 IRF640 can handle 50Wrms for the tests
if it will fail, the only options are the drivers...but, if i want replace them, i've to buy some substitutes... Drivers are: http://www.datasheetcatalog.org/datasheets/50/240361_DS.pdf
Can i use 2SA1837 + 2SC4793??? Hfe is similar, those have higher Vceo but less Ic (1A vs 1.5A). Same pinout. They are drivers from Audison HV16
http://www.datasheetcatalog.org/datasheet/stmicroelectronics/3007.pdf
When it fails, after playing 10 minutes on subwoofer at high volume @1ohm (Re audio), i think Taiwan = fake
i think it was fake because the package was different to220..but i never check datasheet for confirm...i just buy new IRF640N from IR...
Today, searching for my mosfet collection, i find those IRF640. They are only 4 (10 bought - 6 used).
Looking the datasheet specs, i was wrong...i don't think they're fake (i hope).
The string over mosfet said:
IRF640
CC0FG W
MRC 436
(+ ST logo)
Original P-channel is IRF9640 from IR, now IRF9640PbF from Vishay.
Today i can try those 4 IRF640 by ST @4ohm @low power, just for try.
I think 4 IRF640 can handle 50Wrms for the tests
if it will fail, the only options are the drivers...but, if i want replace them, i've to buy some substitutes... Drivers are: http://www.datasheetcatalog.org/datasheets/50/240361_DS.pdf
Can i use 2SA1837 + 2SC4793??? Hfe is similar, those have higher Vceo but less Ic (1A vs 1.5A). Same pinout. They are drivers from Audison HV16
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.