There is no pot of gold at the end of this. Perhaps the journey is not much reward either but I shall try to keep it mildly entertaining.
OVERTURE AND BUGGERUP.
Our quietest member is building a power amp and asked for advice by PM... My phone line/adsl was dead so a barrage of SMSs were exchanged. He might have expected it, ludo writes them 12 deep.
His idea was to build the RAS240 amp as Jannas did at
http://www.avforums.co.za/index.php/topic,20559.0.html
His concern was that he had a PSU with +/-60V rails, in stead of the +/-55V max indicated on his schematic. I suggested that it might work but that an extra pair of fets would be a good idea. Not so simple with a board laid out for two pairs but for first switch-on one would start with only one N channel and one P channel Mosfet in any event. One could then use two pairs for testing and a quick listen to see how hot things got with the elevated supply. By then one would know if further investment is warranted.
Of course there are always other nagging suspicions when one ups the voltage. With the simple resistive current source on the input pair and no degeneration on that pair, at some point one must head off into instability as the current in the input pair gets too high. Still the extra 5V on the positive side is only going to give some 10% increase in current in the input pair which could be on a par with the resistor tolerance. The voltage amplifying stage (VAS) that follows has been compensated in a rather murderous fashion in this amp so all should be well on that front. (I realised later that the board and the schematic are not the same. There is some degeneration for the input pair on the PCB.)
The remaining worry I had was that BD139/140 are 80Volt max parts. In the VAS the voltage can swing pretty close to the rails and even with +/-55V(110V) rails one is sure to exceed the max rating of these transistors at full tilt. To add to the confusion of ifs and maybes, the modern BD139/140 is nothing like the original Philips part. Especially if the modern part was bought from Communica. For all we know those particular BD139/140s are some other high-ish voltage transistors that have been given an appropriate number to match what had been ordered by Communica from the cheapest source they could find. Their BD139/140s have worked reasonably well for me in the past but appeared not to have the high Ft that one would expect. Their origins could be debatable, and their voltage ratings too. They might even stand off more voltage than one would expect.
If one isn't running the amp to clip hard on the rails, transistors rated at well over half the supply voltage should not break when switching the amp on though.
So he takes a deep breath and he fires the thing up. Then he enquires whether the extreme heat he now feels in his spot is to be expected for this design.
Hot? How Hot?
Fookin hot!
In the shimmer he can still see the multimeter. Thers's a 32V offset on the output and she's a-cookin.
AFTERMATH, REMEMBRANCE.
He's hardly started and it's time for the post mortem. I send him the details of how to test the fets. The 2SJ162s measure good but his measurements for the 2SK1058s make no sense at all. They're not dead but they don't work either. Something very fishy there. Natural causes? Not likely.
Now things are difficult to diagnose a thousand miles away and they get particularly hairy as the supply voltage goes up and the threat of fake parts rears it's head. So eventually he mails the populated PCBs off to me. Not the big toroid though.
My phone is still dead at this point. On top of that the PO is on strike, but we have a tracking number and the mama at the PO surely remembers me from our previous incident. The parcel gets to Jhb very quickly and stays there... and stays there.
After some days the intrepid Mr I sends a message that the package is at my local PO in Pta, not in Jhb any more. He has obviously charmed the poppie from Speed Services, so he got inside info that was never entered on the tracking system that the rest of the world uses.
I collect the package, painlessly, from mama. Thank you most kindly ma'am. You are, in fact, a star.
The 2SK1058s are obvious fakes and will never work. See
http://www.avforums.co.za/index.php/topic,23285.0.html
While the boards were in the mail I could not waste all my time browsing the net by reason of the dead phone line, so I'd been busy. There is a new layout for three pairs of mosfets, same size board as the original one. It gets built with real parts and it works right away. I've taken the liberty of sticking to the original Hitachi Application Note somewhat more closely and not doing it exactly like Nico Ras had it. Some sims indicated that other high voltage parts (KSC3503) will probably also do this job better than mystery parts from Communica.
The whole thing does leave a bad taste in the mouth though.
I have to digress into some of the pre-history of all this trouble, so that you may understand. This failure and the effort around it has in fact been some decades in the making. Premorbidity, you might call it, but perhaps it is merely a lack of inventiveness and a distaste for variety on the part of the Cosmos.
Back in the day there was a company that made PA amplifiers locally. Hybrid Electronics in Booysens/Jhb produced the inimitable "Hybrid Titan" in a variety of power levels. If one knew the wrong people, one could make a fair living by repairing those Hybrid Titans, in all of those power levels, and I did. Every week, the same thing. So I got to know these Titanic amps and learnt the seven steps to keep them from constantly coming back:
1) It's always full of roaches. It's been around restaurant kitchens and perhaps even within reach of the buffet. Enough roaches will clog the cooling fan. Spray the infernal emissaries with insecticide.
2) Do the wiring over and do it right. Where those wires run matters, not just where they begin and end. This is not audiophoolery, it's life and death in an amp with power supply wires that are half a meter in length between PSU caps and output stage. There are loops and there are other loops. You get the one loop to make a loop through the other loop and if the fan can still turn it will do the proverbial uneven distribution of the flying what-have-you.
3) Replace the cooked 2SK1058s in the output stage. (The 2SJ162s only die very rarely.) Deburr the heat sink if necessary. There was no need for Mica washers as the heatsinks had the output voltage on them and were well insulated from each other and from the case, but things still have to be flat to get the heat out of the fets and into the extrusion.
4) Overcompensate. No gentlemen, it's nothing at all to do with size. There is a control/feedback loop here and it has to be compensated to remain extremely stable at all frequencies. PA amps are often used to drive speakers with xovers designed
expertly and exclusively by thumbsucking and hearsay. Nobody wonders about the wayward impedances involved. A life in music is about loudness and size and the three B's after all, not silly little technicalities. So for a PA amp stability when driving essentially broken speakers means longevity. Any tendency to oscillation will mean death before sunrise.
5) Change the bias pot so that you can actually bias the mosfets on. For many years, Hybrid used a pair of 200 Ohm preset pots on their driver card. One was supposed to set offset voltage on the output, the other was supposed to set bias. Curiously, the latter 200 Ohm pot was invariably of too low a value to turn any of those fets on to the 80mA of idling current one would expect. On the driver card this pot was also placed in parallel to a fixed resistor. A good safety measure with cheap pots, but in these amps also often a guarantee of near-class-C operation. Some Hybrid amps had all of 1mA output bias, some had less. Whoever set the bias pot during production might have quietly wondered what (s)he was doing wrong for so many, many years, without ever coming across an answer.
6) Fake a star earth by cutting sections out of the track between the main PSU caps and adding some wire. Attatch all the grounds except the transformer centre tap to the fake star, in stead of in a row between the PSU caps.
7) Add high voltage pF value caps to the output devices between Gate and Drain. 47pF/500V/C0G became the stock in trade here when I found a bag full at Electrocomp.
When the Seven Steps had been taken, the amp was reborn.
Once some of the BB&B budget had grudgingly been handed over, even the most deaf of musicians would comment that it sounded a lot better than when it was new.
In spite of considerable efforts on my part, Hybrid remained the reason I avoided live music events unless close friends were involved that I could not appear to snub.
Hence the bad taste. Here I am, making a Hybrid Titan from scratch. It is the same circuit. There's something wrong about it and something uneasy. Like receiving call-up papers for conscription into the SADF again. Or meeting Meneer Maree the History teacher from primary school once more in a dream. One is older and the terror is not the same now, but the neck still tenses up and the reasons for his approach still want clarification.
In the interim, Hybrid, who are still going strong to this day (bearing testimony to the wonderful discretion and a widespread and inherent excellence within the SA Music Industry) have stopped this nonsense. They now distribute something considerably more powerful and much less reliable straight from China. So maybe in the good old days they were pretty bad okes, but still better than now.
(Edit: Fixed Typo in Title)
OVERTURE AND BUGGERUP.
Our quietest member is building a power amp and asked for advice by PM... My phone line/adsl was dead so a barrage of SMSs were exchanged. He might have expected it, ludo writes them 12 deep.
His idea was to build the RAS240 amp as Jannas did at
http://www.avforums.co.za/index.php/topic,20559.0.html
His concern was that he had a PSU with +/-60V rails, in stead of the +/-55V max indicated on his schematic. I suggested that it might work but that an extra pair of fets would be a good idea. Not so simple with a board laid out for two pairs but for first switch-on one would start with only one N channel and one P channel Mosfet in any event. One could then use two pairs for testing and a quick listen to see how hot things got with the elevated supply. By then one would know if further investment is warranted.
Of course there are always other nagging suspicions when one ups the voltage. With the simple resistive current source on the input pair and no degeneration on that pair, at some point one must head off into instability as the current in the input pair gets too high. Still the extra 5V on the positive side is only going to give some 10% increase in current in the input pair which could be on a par with the resistor tolerance. The voltage amplifying stage (VAS) that follows has been compensated in a rather murderous fashion in this amp so all should be well on that front. (I realised later that the board and the schematic are not the same. There is some degeneration for the input pair on the PCB.)
The remaining worry I had was that BD139/140 are 80Volt max parts. In the VAS the voltage can swing pretty close to the rails and even with +/-55V(110V) rails one is sure to exceed the max rating of these transistors at full tilt. To add to the confusion of ifs and maybes, the modern BD139/140 is nothing like the original Philips part. Especially if the modern part was bought from Communica. For all we know those particular BD139/140s are some other high-ish voltage transistors that have been given an appropriate number to match what had been ordered by Communica from the cheapest source they could find. Their BD139/140s have worked reasonably well for me in the past but appeared not to have the high Ft that one would expect. Their origins could be debatable, and their voltage ratings too. They might even stand off more voltage than one would expect.
If one isn't running the amp to clip hard on the rails, transistors rated at well over half the supply voltage should not break when switching the amp on though.
So he takes a deep breath and he fires the thing up. Then he enquires whether the extreme heat he now feels in his spot is to be expected for this design.
Hot? How Hot?
Fookin hot!
In the shimmer he can still see the multimeter. Thers's a 32V offset on the output and she's a-cookin.
AFTERMATH, REMEMBRANCE.
He's hardly started and it's time for the post mortem. I send him the details of how to test the fets. The 2SJ162s measure good but his measurements for the 2SK1058s make no sense at all. They're not dead but they don't work either. Something very fishy there. Natural causes? Not likely.
Now things are difficult to diagnose a thousand miles away and they get particularly hairy as the supply voltage goes up and the threat of fake parts rears it's head. So eventually he mails the populated PCBs off to me. Not the big toroid though.
My phone is still dead at this point. On top of that the PO is on strike, but we have a tracking number and the mama at the PO surely remembers me from our previous incident. The parcel gets to Jhb very quickly and stays there... and stays there.
After some days the intrepid Mr I sends a message that the package is at my local PO in Pta, not in Jhb any more. He has obviously charmed the poppie from Speed Services, so he got inside info that was never entered on the tracking system that the rest of the world uses.
I collect the package, painlessly, from mama. Thank you most kindly ma'am. You are, in fact, a star.
The 2SK1058s are obvious fakes and will never work. See
http://www.avforums.co.za/index.php/topic,23285.0.html
While the boards were in the mail I could not waste all my time browsing the net by reason of the dead phone line, so I'd been busy. There is a new layout for three pairs of mosfets, same size board as the original one. It gets built with real parts and it works right away. I've taken the liberty of sticking to the original Hitachi Application Note somewhat more closely and not doing it exactly like Nico Ras had it. Some sims indicated that other high voltage parts (KSC3503) will probably also do this job better than mystery parts from Communica.
The whole thing does leave a bad taste in the mouth though.
I have to digress into some of the pre-history of all this trouble, so that you may understand. This failure and the effort around it has in fact been some decades in the making. Premorbidity, you might call it, but perhaps it is merely a lack of inventiveness and a distaste for variety on the part of the Cosmos.
*********
Back in the day there was a company that made PA amplifiers locally. Hybrid Electronics in Booysens/Jhb produced the inimitable "Hybrid Titan" in a variety of power levels. If one knew the wrong people, one could make a fair living by repairing those Hybrid Titans, in all of those power levels, and I did. Every week, the same thing. So I got to know these Titanic amps and learnt the seven steps to keep them from constantly coming back:
1) It's always full of roaches. It's been around restaurant kitchens and perhaps even within reach of the buffet. Enough roaches will clog the cooling fan. Spray the infernal emissaries with insecticide.
2) Do the wiring over and do it right. Where those wires run matters, not just where they begin and end. This is not audiophoolery, it's life and death in an amp with power supply wires that are half a meter in length between PSU caps and output stage. There are loops and there are other loops. You get the one loop to make a loop through the other loop and if the fan can still turn it will do the proverbial uneven distribution of the flying what-have-you.
3) Replace the cooked 2SK1058s in the output stage. (The 2SJ162s only die very rarely.) Deburr the heat sink if necessary. There was no need for Mica washers as the heatsinks had the output voltage on them and were well insulated from each other and from the case, but things still have to be flat to get the heat out of the fets and into the extrusion.
4) Overcompensate. No gentlemen, it's nothing at all to do with size. There is a control/feedback loop here and it has to be compensated to remain extremely stable at all frequencies. PA amps are often used to drive speakers with xovers designed
expertly and exclusively by thumbsucking and hearsay. Nobody wonders about the wayward impedances involved. A life in music is about loudness and size and the three B's after all, not silly little technicalities. So for a PA amp stability when driving essentially broken speakers means longevity. Any tendency to oscillation will mean death before sunrise.
5) Change the bias pot so that you can actually bias the mosfets on. For many years, Hybrid used a pair of 200 Ohm preset pots on their driver card. One was supposed to set offset voltage on the output, the other was supposed to set bias. Curiously, the latter 200 Ohm pot was invariably of too low a value to turn any of those fets on to the 80mA of idling current one would expect. On the driver card this pot was also placed in parallel to a fixed resistor. A good safety measure with cheap pots, but in these amps also often a guarantee of near-class-C operation. Some Hybrid amps had all of 1mA output bias, some had less. Whoever set the bias pot during production might have quietly wondered what (s)he was doing wrong for so many, many years, without ever coming across an answer.
6) Fake a star earth by cutting sections out of the track between the main PSU caps and adding some wire. Attatch all the grounds except the transformer centre tap to the fake star, in stead of in a row between the PSU caps.
7) Add high voltage pF value caps to the output devices between Gate and Drain. 47pF/500V/C0G became the stock in trade here when I found a bag full at Electrocomp.
When the Seven Steps had been taken, the amp was reborn.
Once some of the BB&B budget had grudgingly been handed over, even the most deaf of musicians would comment that it sounded a lot better than when it was new.
In spite of considerable efforts on my part, Hybrid remained the reason I avoided live music events unless close friends were involved that I could not appear to snub.
Hence the bad taste. Here I am, making a Hybrid Titan from scratch. It is the same circuit. There's something wrong about it and something uneasy. Like receiving call-up papers for conscription into the SADF again. Or meeting Meneer Maree the History teacher from primary school once more in a dream. One is older and the terror is not the same now, but the neck still tenses up and the reasons for his approach still want clarification.
In the interim, Hybrid, who are still going strong to this day (bearing testimony to the wonderful discretion and a widespread and inherent excellence within the SA Music Industry) have stopped this nonsense. They now distribute something considerably more powerful and much less reliable straight from China. So maybe in the good old days they were pretty bad okes, but still better than now.
*******
(Edit: Fixed Typo in Title)
