Well I've run some more tests and I still have some parts to check. I am now ready to remove items from the board to test each one. Once again though I am hoping for some direction. With one channel completely connected and P1 set for max I placed F4 in the holder and placed the amp meter across F2's place. Black lead on capacitor side - Yes I used the light bulb tester first. Reading on meter is about -7.5 (meter set to 200m). Of course P1 has no effect on this reading. Curiosity got the better of me so I moved the fuse to F2 and placed the leads over F4. Meter reads about 45 ( i didn't bother even adjusting P1 at this point). The only part that is different from the recommended parts is C6. I have a 330 bipolar 16v cap with a jumper in C6b. I belive the plans call for a 220 bipolar cap.
I also noticed the led on the power switch stayed lit after I turned it off. It is powered from the cap side of F2-F3 (+). I do not believe this happend when I was testing the other channel durring the previous posts.
As always, anything to watch out for is greatly appreciated.
Steve
I also noticed the led on the power switch stayed lit after I turned it off. It is powered from the cap side of F2-F3 (+). I do not believe this happend when I was testing the other channel durring the previous posts.
As always, anything to watch out for is greatly appreciated.
Steve
Hi,
If my memory is correct (about the quiescent currents), then you need to look at volts drop in the voltage amplifier section. Do not remove anything just yet.
Go back to the schematic and calculate the current and voltage across every resistor. Then compare these values against your amp.
If my memory is correct (about the quiescent currents), then you need to look at volts drop in the voltage amplifier section. Do not remove anything just yet.
Go back to the schematic and calculate the current and voltage across every resistor. Then compare these values against your amp.
Hi,
why did you change the 200uF to 330uF?
The 1k0 resistor gave F-3db @ 0.7Hz using 220uF. A bit low considering you will have a high pass filter somewhere else.
330uF drops this down to 0.5Hz and for no benefit.
You could fit either 120uF or 150uF (or a series pair of 270uF) and still get exemplary sub-bass performance.
D.Self says measure the AC voltage across this cap and ensure that at maximum output voltage it develops a low AC voltage to ensure there is minimum distortion caused by this cap. RC=140mS gets one into the right region to pass this test.
why did you change the 200uF to 330uF?
The 1k0 resistor gave F-3db @ 0.7Hz using 220uF. A bit low considering you will have a high pass filter somewhere else.
330uF drops this down to 0.5Hz and for no benefit.
You could fit either 120uF or 150uF (or a series pair of 270uF) and still get exemplary sub-bass performance.
D.Self says measure the AC voltage across this cap and ensure that at maximum output voltage it develops a low AC voltage to ensure there is minimum distortion caused by this cap. RC=140mS gets one into the right region to pass this test.
Q7 may be shorted, Check the voltage across across the CE junction of Q7.
Place a voltmeter between the base of Q14 and the base of Q15. P1 should vary this voltage from nearly zero to 3.x volts. This is an direct indication that the bias regulator is working. At some point the voltage starts exceeding the Vbe drops of the output transistors and the transistors start to conduct, placing the amplifier in class AB mode.
If the voltage between the base of Q14 and Q15 is less than where the transistors start to conduct, you have a class B amplifier and crossover distortion.
Nearly 30 years ago, I troublesho0t the bias regulator with a small power supply that had a 0-20 mA and 0-200 mA current meter and went to about 20 volts.
You might be able to troubleshoot the regulator without +-50 V power by injecting a current limited supply beween the juction of D9, C10 (+) and D8, C11 (-). Thinking something like 9V battery in series with about 500 ohms. P1 should vary the voltage across C12.
Place a voltmeter between the base of Q14 and the base of Q15. P1 should vary this voltage from nearly zero to 3.x volts. This is an direct indication that the bias regulator is working. At some point the voltage starts exceeding the Vbe drops of the output transistors and the transistors start to conduct, placing the amplifier in class AB mode.
If the voltage between the base of Q14 and Q15 is less than where the transistors start to conduct, you have a class B amplifier and crossover distortion.
Nearly 30 years ago, I troublesho0t the bias regulator with a small power supply that had a 0-20 mA and 0-200 mA current meter and went to about 20 volts.
You might be able to troubleshoot the regulator without +-50 V power by injecting a current limited supply beween the juction of D9, C10 (+) and D8, C11 (-). Thinking something like 9V battery in series with about 500 ohms. P1 should vary the voltage across C12.
Update and advice for all first timers!
After some testing as directed above (thanks again) I came to the conclusion that there was more then one blown item on the board. I then did what I should have done from the very start. (This is the advice part) I started fresh and checked each component’s value and orientation before placing it on the board. The multimeter with transistor and capacitor checking functions was invaluable here. Checked configuration of power transistors and diodes on the heat sinks and confirmed all wiring from the heat sinks to the board. Checked for any short circuits as described on the professors web site. Placed 100ohm resister across F2 and F4 and used a light bulb tester inline with the hot. Power up was good. Checked some voltages at various places. Just to be safe I powered down and increased the wattage of the bulb. Power up went smooth. Lastly, I removed the light bulb tester, and the 100ohm resistors. Placed the fuse in F4 and DMM reading DV amps in place of F2. Power up was good. Adjusted P1 to get close to 100mA (took about 20 minutes as that pot seems to have a real hair trigger). Anyway the amps seem to float between 99.9 and 100.4. I read 3.44volts across the collectors of Q12 and Q13.
All my problems seem to have arisen because of the incorrect resistor as detailed earlier. Consider it –Lesson Learned.
Once again, thanks to all who gave advice and direction. I will be completing the above procedure with the other channel over the next couple of days. Do a little more testing, and then hopefully, I will be able to have a listen.
Curioprop
After some testing as directed above (thanks again) I came to the conclusion that there was more then one blown item on the board. I then did what I should have done from the very start. (This is the advice part) I started fresh and checked each component’s value and orientation before placing it on the board. The multimeter with transistor and capacitor checking functions was invaluable here. Checked configuration of power transistors and diodes on the heat sinks and confirmed all wiring from the heat sinks to the board. Checked for any short circuits as described on the professors web site. Placed 100ohm resister across F2 and F4 and used a light bulb tester inline with the hot. Power up was good. Checked some voltages at various places. Just to be safe I powered down and increased the wattage of the bulb. Power up went smooth. Lastly, I removed the light bulb tester, and the 100ohm resistors. Placed the fuse in F4 and DMM reading DV amps in place of F2. Power up was good. Adjusted P1 to get close to 100mA (took about 20 minutes as that pot seems to have a real hair trigger). Anyway the amps seem to float between 99.9 and 100.4. I read 3.44volts across the collectors of Q12 and Q13.
All my problems seem to have arisen because of the incorrect resistor as detailed earlier. Consider it –Lesson Learned.
Once again, thanks to all who gave advice and direction. I will be completing the above procedure with the other channel over the next couple of days. Do a little more testing, and then hopefully, I will be able to have a listen.
Curioprop
output bias current drops on warming up
Hi,
The Leaches I've built seem to be overcompensated.
As they warm up the bias current drops.
One has SMD diodes and the other wire ended. 4off in both cases.
Other than removing/shorting out one diode, is there a method to reduce the temperature compensation?
Hi,
The Leaches I've built seem to be overcompensated.
As they warm up the bias current drops.
One has SMD diodes and the other wire ended. 4off in both cases.
Other than removing/shorting out one diode, is there a method to reduce the temperature compensation?
Once again, thanks to all who replied. The amp is together, powered up and sounding great. I have a lot of CD's to go though (and albums).
The speakers are getting a good work over.
Thanks again.
Of course I am now planing the next project. I am thinking of the ESP P88 Preamp. Anyways, thats for another thread.
Curioprop
The speakers are getting a good work over.
Thanks again.
Of course I am now planing the next project. I am thinking of the ESP P88 Preamp. Anyways, thats for another thread.
Curioprop
Hi,
I have just completed some full power testing (report in the main thread).
the heatsink was lying flat rather than upright and as expected it got pretty hot, I guess probably as hot as it will ever get in normal operating conditions.
I have not modified the temperature compensation from the normal 4diodes in close contact with the output heatsink. The drivers and pre-drivers have their own cooling.
The Vre bias voltage fell from 25mV to 15mV, using up the whole of the recommended range for optimum ClassAB bias. In that repect it works.
If the drivers were on the output heatsink the over-compensation would be much worse.
I have just completed some full power testing (report in the main thread).
the heatsink was lying flat rather than upright and as expected it got pretty hot, I guess probably as hot as it will ever get in normal operating conditions.
I have not modified the temperature compensation from the normal 4diodes in close contact with the output heatsink. The drivers and pre-drivers have their own cooling.
The Vre bias voltage fell from 25mV to 15mV, using up the whole of the recommended range for optimum ClassAB bias. In that repect it works.
If the drivers were on the output heatsink the over-compensation would be much worse.
My heatsinks (original Leach) are mounted flat and I have the 3 diode version of compensation, BUT I probably have the Kapton isulators rater than Mica. Some Figures of merit:
Silicon/Fiberglass 0.5 deg C/W
Kapton 0.4 deg C/W
Mica 0.009 deg C/W (bad)
How well it's compensated may indeed depend on the type of insulator used.
Two of the diodes are a really tight fit through holes in the heatsink and has a plastic body. The other lies flat across the heatsink. In normal operation, heatsink is barely warm. I haven't done full power tests in a great while.
Silicon/Fiberglass 0.5 deg C/W
Kapton 0.4 deg C/W
Mica 0.009 deg C/W (bad)
How well it's compensated may indeed depend on the type of insulator used.
Two of the diodes are a really tight fit through holes in the heatsink and has a plastic body. The other lies flat across the heatsink. In normal operation, heatsink is barely warm. I haven't done full power tests in a great while.
Hmm. I just put together a new Leach amplifier using one of Brian Bell's boards and MJL21193/94 outputs mounted on mica insulators. The heatsink is mounted with the fins vertical but it is smaller than I'd like and biased for lowest distortion gets too hot to touch.
Looks like I've got some silicon/fiberglass insulators, will have to try those. I may have to look for some Kapton insulators, too.
If I can get a second board from Brian, I will try the set of NJL21193/94 output devices that have the integrated thermal diodes. Thus far, I have had no problems with high temperature and bias stability, really don't expect any. But I haven't really tried to abuse the amplifier.
Yet. (heh, heh, heh...)
(Note that OnSemi has Thermaltrak devices in both the -21193/94 and -1302/3281 flavors (the latter have faster Ft's.)
Distortion measures less than .008% THD at low power levels, but climbs to .1% at higher power and frequencies. I'm seeing a bit more high order distortion products than I think I would like to see at higher power. With plenty of bias at low power level distortion disappears into the noise. A 1 uF capacitor across the driver emitters makes a _tiny_ bit of difference in distortion, but only at high frequencies. It's probably not a worthwhile mod to incorporate, but doesn't appear to do anything harmful. No attempt to match output devices was made, though I did try to match the diff amp pairs. I will try several different IM distortion measurements in my next test session.
The amplifier was formerly a Dove Systems Dimmermaster theatre light control system that I picked up at Boeing Surplus for $35. It makes for a nice compact amplifier chassis and I'm rather pleased with it. Solid black and looks a lot like many of the low profile Hafler amplifiers. First Leach amp I've built that actually >looks< good.
Looks like I've got some silicon/fiberglass insulators, will have to try those. I may have to look for some Kapton insulators, too.
If I can get a second board from Brian, I will try the set of NJL21193/94 output devices that have the integrated thermal diodes. Thus far, I have had no problems with high temperature and bias stability, really don't expect any. But I haven't really tried to abuse the amplifier.
Yet. (heh, heh, heh...)
(Note that OnSemi has Thermaltrak devices in both the -21193/94 and -1302/3281 flavors (the latter have faster Ft's.)
Distortion measures less than .008% THD at low power levels, but climbs to .1% at higher power and frequencies. I'm seeing a bit more high order distortion products than I think I would like to see at higher power. With plenty of bias at low power level distortion disappears into the noise. A 1 uF capacitor across the driver emitters makes a _tiny_ bit of difference in distortion, but only at high frequencies. It's probably not a worthwhile mod to incorporate, but doesn't appear to do anything harmful. No attempt to match output devices was made, though I did try to match the diff amp pairs. I will try several different IM distortion measurements in my next test session.
The amplifier was formerly a Dove Systems Dimmermaster theatre light control system that I picked up at Boeing Surplus for $35. It makes for a nice compact amplifier chassis and I'm rather pleased with it. Solid black and looks a lot like many of the low profile Hafler amplifiers. First Leach amp I've built that actually >looks< good.
I think your heatsinking logic is back to front.Damon Hill said:
using a poor thermal conductor will increase the semi case temperature and REDUCE the heatsink temperature.
Using a good thermal conductor will decrease the semi case temperature and INCREASE the heatsink temperature.
If your sink already runs too hot and you improve the thermal conductivity of the electrical insulator then you will increase the sink temperature very slightly. But there is a very big gain in reduced semi case temperature and this can improve reliability.
Thin mica insulators applied with thermal compound to both sides comes roughly in the middle of the thermal conductivity range of the dry fit insulators. You need to select alternatives carefully to ensure you get what you expect. The good thermal conductors tend to be very expensive the poor ones are just expensive.
Are you running two pair 21193/4? I would have matched Vbe at intended output bias current and matched Re to ensure any measurements taken later could be relied on.
I really didn't have a choice about matching; I have sample sets in small numbers and made only a cursory check. They're from the same date code groups, and it seemed that those groups tracked reasonably well; PNP/NPN matching does not appear practical and having imperfectly matched sets never seemed to be a problem other than distortion possibly being higher. I may try matching more closely in the future when I have more devices to choose among. Just looking at the bias current across the emitter resistors doesn't suggest any serious mismatches, and sustained operation at full power didn't provoke any problems. I'm running in pairs, with 52 volt rails until I can get properly rated power transformers (or maybe not).
I may have misunderstood the values posted for the different types of insulators (I had been thinking mica was the most thermally conductive); I'd rather go with the most thermally conductive and let the heatsink get as hot as it will.
Mounting the Thermaltrak devices will be fun as the emitter pin will have to be connected by jumper and the diode connectors by jumpers as well. A bit messy but doable; it will be interesting to see how bias responds to temperature changes compared to the diode string I expoxied to the heatsink. Is there a high-temperature tolerant epoxy?
I may have misunderstood the values posted for the different types of insulators (I had been thinking mica was the most thermally conductive); I'd rather go with the most thermally conductive and let the heatsink get as hot as it will.
Mounting the Thermaltrak devices will be fun as the emitter pin will have to be connected by jumper and the diode connectors by jumpers as well. A bit messy but doable; it will be interesting to see how bias responds to temperature changes compared to the diode string I expoxied to the heatsink. Is there a high-temperature tolerant epoxy?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.