Hugh
Thanks for the quick responce
I think we can use the same devices for VAS and predriver stages ,
I had a 70mhz device in mind , I can also use the devices recomended
by you if they are avilible ....
You asked me about the emitter resistor of the output devices , I have
decided to use 0.25 ohms , as on higher value the voltage drop at peak
current 5.5A will be too much , however there is a base resistor of 10
ohms with every output device which along with the emitter resistance
will help to maintain the current sharing of the output devices .
Please do not scare me |||||||||
" If you want this thing to sound good you will need same batch output
device matching (this reduces sample size to perhaps 100 for 8 exactly
alike devices), and very careful voicing of all operating points and
capacitors in the circuit for best sonics."
I cannot afford to select 8 matched devices out of 100 but I will select 8
out of 25 , I beleive nowadays these new devices are having very little
differances in their beta ????
" If the former, are you able to control the offset properly? Remember,
high voltages here, and we are juggling low impedance power supplies
which eat voice coils for breakfast!! "
Yes the ofset will be controlled properly -- I hope ?? I do not want the
power supplies to eat voice coils for breakfast!!!! Ha ha ,
loudspeakers are expensive too .....
The peak output power at 2ohms will be 3784w and if we calculate max
peak power dissipated by the output devices @ 50% then each output
device will dissipate 118w peak. is this OK ???
The peak power dissipation of the driver will be 24w , this means I will
have to mount the driver with the output devices on the main heatsink ??
Thanks for saying that " It's all doable "
I will do it
Rajeev
Thanks for the quick responce
I think we can use the same devices for VAS and predriver stages ,
I had a 70mhz device in mind , I can also use the devices recomended
by you if they are avilible ....
You asked me about the emitter resistor of the output devices , I have
decided to use 0.25 ohms , as on higher value the voltage drop at peak
current 5.5A will be too much , however there is a base resistor of 10
ohms with every output device which along with the emitter resistance
will help to maintain the current sharing of the output devices .
Please do not scare me |||||||||
" If you want this thing to sound good you will need same batch output
device matching (this reduces sample size to perhaps 100 for 8 exactly
alike devices), and very careful voicing of all operating points and
capacitors in the circuit for best sonics."
I cannot afford to select 8 matched devices out of 100 but I will select 8
out of 25 , I beleive nowadays these new devices are having very little
differances in their beta ????
" If the former, are you able to control the offset properly? Remember,
high voltages here, and we are juggling low impedance power supplies
which eat voice coils for breakfast!! "
Yes the ofset will be controlled properly -- I hope ?? I do not want the
power supplies to eat voice coils for breakfast!!!! Ha ha ,
loudspeakers are expensive too .....
The peak output power at 2ohms will be 3784w and if we calculate max
peak power dissipated by the output devices @ 50% then each output
device will dissipate 118w peak. is this OK ???
The peak power dissipation of the driver will be 24w , this means I will
have to mount the driver with the output devices on the main heatsink ??
Thanks for saying that " It's all doable "
I will do it
Rajeev
Hold your horses.
#1 The maximum loss power declared in all specs is a theoretircal one, read it carefully and see that this power is at chip temperature of 25 degrees C. Then there is derate at temperatures higher than that.
#2 And even more annoying is SECOND BREAKDOWN a horrifying term, that I remarkably not seen discussed in this forum. This IS the primary fault condition. Time again to read past the first page of the specs. A little curve where the maximums of the transistor appears as lines. there is one horizontal line depicting the max Vce, a vertical line indicating the melt down current of bondwires Ic max, there is one straigh line (due to logaritmic scales) of maximum power dissipation as mentioned above, THEN there is a mystical break in this line that indicates that by some mysterious ways of faith in this area, most often at Vce > 30V up to Vce max the transistor can't handle maximum power. This is the second breakdown line. Look for yourselves, obey this line even though it looks tough. Because if your amp is able to step over this line the transistor in it's linear operation, as we are dealing with here, is squezing all the current through smaller and smaller tunnels in the silicon the current density in these very narrow channels wil melt the silicon. This can lead to leakcurrents but mostly to a catastrophe.
#3 Don't pay som much attention to max current etc. There is a magic rule that I learned from MOTOROLA (The King of useful transistors, the rest is CRAP when it comes durability).
Calculate Imax=Umax/Rload.
You won't be cheating nobody but yourself.
Check the curve mentioned before, that these points are safely within the curve.
Imax and Umax/2
Imax/2 and Umax
Especially the second can be tough. There is, I believe no transitor can take 100V and more than 1A at the same time.
There might be some exclusions from this but they are rare.
From a guy NOT needing or using relays at the output.
Andreas W
PS These relays are a pain anyhow, and at Vcc over 60 V they don't help you much, since they melt so nicely. DS
#1 The maximum loss power declared in all specs is a theoretircal one, read it carefully and see that this power is at chip temperature of 25 degrees C. Then there is derate at temperatures higher than that.
#2 And even more annoying is SECOND BREAKDOWN a horrifying term, that I remarkably not seen discussed in this forum. This IS the primary fault condition. Time again to read past the first page of the specs. A little curve where the maximums of the transistor appears as lines. there is one horizontal line depicting the max Vce, a vertical line indicating the melt down current of bondwires Ic max, there is one straigh line (due to logaritmic scales) of maximum power dissipation as mentioned above, THEN there is a mystical break in this line that indicates that by some mysterious ways of faith in this area, most often at Vce > 30V up to Vce max the transistor can't handle maximum power. This is the second breakdown line. Look for yourselves, obey this line even though it looks tough. Because if your amp is able to step over this line the transistor in it's linear operation, as we are dealing with here, is squezing all the current through smaller and smaller tunnels in the silicon the current density in these very narrow channels wil melt the silicon. This can lead to leakcurrents but mostly to a catastrophe.
#3 Don't pay som much attention to max current etc. There is a magic rule that I learned from MOTOROLA (The King of useful transistors, the rest is CRAP when it comes durability).
Calculate Imax=Umax/Rload.
You won't be cheating nobody but yourself.
Check the curve mentioned before, that these points are safely within the curve.
Imax and Umax/2
Imax/2 and Umax
Especially the second can be tough. There is, I believe no transitor can take 100V and more than 1A at the same time.
There might be some exclusions from this but they are rare.
From a guy NOT needing or using relays at the output.
Andreas W
PS These relays are a pain anyhow, and at Vcc over 60 V they don't help you much, since they melt so nicely. DS
Andreas W said:
Which is why it's a good idea to use DPDT relays in the supply rails rather than the output to disable the amp. in the event of a device failure...
Hi Andreas,
You normally want even more headroom on an output stage to cover some fault conditions. Every tough amp out there has a very conservative rating on heatsinks and output current capability. Second breakdown is something many amps die from. Normally in a fault condition.
You always need some way of removing energy from a load when an abnormal condition exists. I don't care how you do it, it just has to be done. Your designs do need protection, as every other design requires it. Components can fail "just 'cause". This may create a large DC offset.
Years ago, the guys at Bryston told me that their amps never fail. He meant never ever. I wonder where they get the magic components they use. I have repaired dead Brystons that were idling when they died. Never believe that an amp design is so special that it is immune to the failure rates pulished by component manufacturers.
You always need a way to automatically disconnect a load (speaker here) in the event of a fault in the source. Period.
I use speaker relays. This is just as valid as any other method as speaker fuses have their own problems. BTW, use the correct relay and they do not "melt".
-Chris
You normally want even more headroom on an output stage to cover some fault conditions. Every tough amp out there has a very conservative rating on heatsinks and output current capability. Second breakdown is something many amps die from. Normally in a fault condition.
You always need some way of removing energy from a load when an abnormal condition exists. I don't care how you do it, it just has to be done. Your designs do need protection, as every other design requires it. Components can fail "just 'cause". This may create a large DC offset.
Years ago, the guys at Bryston told me that their amps never fail. He meant never ever. I wonder where they get the magic components they use. I have repaired dead Brystons that were idling when they died. Never believe that an amp design is so special that it is immune to the failure rates pulished by component manufacturers.
You always need a way to automatically disconnect a load (speaker here) in the event of a fault in the source. Period.
I use speaker relays. This is just as valid as any other method as speaker fuses have their own problems. BTW, use the correct relay and they do not "melt".
-Chris
Andreas W said:
Then the rest must be really crappy, because I used Motorola
for 18 years and still had to SOA test each output to be certain.
Eva said:
Digi-key still has C3281/A1302 for sale. Can these be trusted to be the real parts if production stop in 1999? They are a supplier for Toshiba. How can these be fakes?
Sze said:
Digi-Key also has 5200/1943 in stock at great prices... now why would anyone want to take a chance with the 1302/3281?
I was a little tired when i wrote, to fit these points safely within the safe operating area at worst possible conditions is amust. Of course one must have the traditional short protections on the transistors as well otherwise you don't know what is the maximum current at a given voltage.
The relays are a mixed blessing, both in the sense of prottecting and in listening. Since the pressure on the the contacts is ridicuolusly low and they are prone to get a little oxide on them they are not what can be called an ideal connection. At low levels I wouldn't be surprised if they are "audible" and therefore unwanted. I think it is better still to put them on the secondary side of the transformer. There their imperfections are hard to pass through the massive filtering of the electrolytics. Of course the transistors will have to discharge the caps in event of failure.
Poweramplifier giving out thumps of more than a few volts are poor constructions and should be reengineered.
An other perspective of robustness in amplifier construction is to make amplifiers stable under all conditions, ie any voltage and virtually any load. Instabilities under some load conditions is a dangerous situation, HF oscillation can produce tremendous heat. Exterminating any prior margins in heat evaporation.
Any construction should be tested for stability with square wave and a few different loads, like 8ohm || 1uf at 10kHz square wave.
Andreas
The relays are a mixed blessing, both in the sense of prottecting and in listening. Since the pressure on the the contacts is ridicuolusly low and they are prone to get a little oxide on them they are not what can be called an ideal connection. At low levels I wouldn't be surprised if they are "audible" and therefore unwanted. I think it is better still to put them on the secondary side of the transformer. There their imperfections are hard to pass through the massive filtering of the electrolytics. Of course the transistors will have to discharge the caps in event of failure.
Poweramplifier giving out thumps of more than a few volts are poor constructions and should be reengineered.
An other perspective of robustness in amplifier construction is to make amplifiers stable under all conditions, ie any voltage and virtually any load. Instabilities under some load conditions is a dangerous situation, HF oscillation can produce tremendous heat. Exterminating any prior margins in heat evaporation.
Any construction should be tested for stability with square wave and a few different loads, like 8ohm || 1uf at 10kHz square wave.
Andreas
Hi Andreas,
I'm am more concerned about the amp that is designed well and not abused. They can fail due due a component just dying. Speaker relays do need replacing from time to time. Relays in good condition will affect the sound less than a speaker fuse.
I do agree with your point. A poor design is a poor design, no argument. It's amazing how many iffy designs are put into production. Many of these do not have any protecton at all. The Nikko NA-100 come to mind. Bryston 4B's have killed many speakers on the way out too.
-Chris
I'm am more concerned about the amp that is designed well and not abused. They can fail due due a component just dying. Speaker relays do need replacing from time to time. Relays in good condition will affect the sound less than a speaker fuse.
I do agree with your point. A poor design is a poor design, no argument. It's amazing how many iffy designs are put into production. Many of these do not have any protecton at all. The Nikko NA-100 come to mind. Bryston 4B's have killed many speakers on the way out too.
-Chris
Speaker relays wouldn't be such a problem if the contacts were plated with gold or platinum. But would this really be practical??....$$$
Gold and platinum can be arced or burnt just like anything else. All you really need are like metalic contacts in good condition. When the contacts become worn, damaged or oxidized it is time to clean or replace the relay. Not a big deal.
No point in having expensive contacts. If this is a problem, why not recommend the MIL spec'd mercury wetted relay contacts? Aside from the mercury issue.
Nothing lasts forever, all equipment needs maintenance.
-Chris
No point in having expensive contacts. If this is a problem, why not recommend the MIL spec'd mercury wetted relay contacts? Aside from the mercury issue.
Nothing lasts forever, all equipment needs maintenance.
-Chris
anatech said:
True. Now only if Dan'D Agostino had desiged a user servicable relay change board so that the %#^@$% relays on my KSA-250 would be dead quiet.
I doesn't matter what the relay contacts are made of when the amplifier blows and the relay has to interrupt DC, they arc weld.
Same thing happens if the relay is moved to the ±rails.
Use the relay to keep things quiet in normal operation
Use a crowbar for DC protection.
Same thing happens if the relay is moved to the ±rails.
Use the relay to keep things quiet in normal operation
Use a crowbar for DC protection.
Hi djk,
Nothing wrong with that approach. High power amps need a crowbar across the output. The only exception I've seen is a Carver-like supply that stores little energy and can be shut down quickly. Low power amps generally don't weld the contacts of a relay closed, but they do arc. So change the relay in that case.
-Chris
Nothing wrong with that approach. High power amps need a crowbar across the output. The only exception I've seen is a Carver-like supply that stores little energy and can be shut down quickly. Low power amps generally don't weld the contacts of a relay closed, but they do arc. So change the relay in that case.
-Chris
It's quite easy to construct an amp that is quiet at start up and shut down. Just follow this, hook up a variac powersupply or some hefty bench power supply that can supply the needed voltage current is not a big issue here. Crank it up and down a watch the scope on the output while you have some load and a few volts output. Observe when the amp ****ups this could be the input stage or the voltage amplifier, the next stage. If it is the input stage try to invent some new constant current hookup that works at a lower voltage. Same thing goes with the next stage. Add a little cap 0.22 -- 1 mfd or more to the output get some over shoot at the output, with 1 - 10 kHz input, look for more than one oscillation of high frequiency. This is an indication of lacking phasemargin. Adjust the phase margin by increasing the miller or shunt cap controlling this. turn the voltage up and down as to get the amp working under all different possible supply conditions. when there is no undamped HF oscillation, and the output is controlled down to a few volts, the amp will be able to run without a relay. Use a triac with a RC AND a triac firing device a DIAC, otherwise there won't be enough gate current to trigger the triac, which will have to be big, 40amps or so. Follow this and your amp will be reliable and nice behaving. And probably sound better since instabilities sound strange and relay contacts makes the sound jerky and the stereo image gets blurred. Try to solder as much as you can and avoid connectors as much as possible.
Andreas
Andreas
Hi Andreas,
Start up and shutdown muting wasn't my concern. Protection in a failure mode is more important given a stable design. My speakers are worth far more than an amp repair (needed anyhow if there is a problem). I have seen more than one amplifier start a fire in the speaker(s). This is even more important if your equipment is left on 24/7. Another practice I disagree with.
-Chris
Start up and shutdown muting wasn't my concern. Protection in a failure mode is more important given a stable design. My speakers are worth far more than an amp repair (needed anyhow if there is a problem). I have seen more than one amplifier start a fire in the speaker(s). This is even more important if your equipment is left on 24/7. Another practice I disagree with.
-Chris
Re: Arching relay contacts
I remember seeing somewhere, a long time ago (in a place far far away etc..) a power supply where the amplifier rails were fed by high current, low Rds-on MOSFETS. DC detection simply removed the gate drive and turned the MOSFETS off. No arc no fuss. The MOSFETS were after the main filter capacitors and fed a smaller set before the amp.
I never tried it at the time because I was concerned about the reduction of power supply impedance even though the Rds-on was only about 0.02 ohms. I though at the time that with additional circuitry these MOSFETS could provide short cct protection as well.
Would this work or have I missed something.
Cheers
I mean increase in power supply impedance ...oh you know what I mean ...
I remember seeing somewhere, a long time ago (in a place far far away etc..) a power supply where the amplifier rails were fed by high current, low Rds-on MOSFETS. DC detection simply removed the gate drive and turned the MOSFETS off. No arc no fuss. The MOSFETS were after the main filter capacitors and fed a smaller set before the amp.
I never tried it at the time because I was concerned about the reduction of power supply impedance even though the Rds-on was only about 0.02 ohms. I though at the time that with additional circuitry these MOSFETS could provide short cct protection as well.
Would this work or have I missed something.
Cheers
I mean increase in power supply impedance ...oh you know what I mean ...
Hi,
The late John Linsley-Hood did a mosfet output poweramp design, late 80's I think for ETI. It included a discrete regulated power supply. The psu was quite clever, it had two low current hi V rails, two hi current lower volt rails for the output fets, fold back current limiting, dc detect & shutdown, asymetric rails shutdown and maybe some other bits. Oh and the output trs of the psu were also fets.
Andrew T.
The late John Linsley-Hood did a mosfet output poweramp design, late 80's I think for ETI. It included a discrete regulated power supply. The psu was quite clever, it had two low current hi V rails, two hi current lower volt rails for the output fets, fold back current limiting, dc detect & shutdown, asymetric rails shutdown and maybe some other bits. Oh and the output trs of the psu were also fets.
Andrew T.
The extra higher voltage supplies on the MOSFET power supplies where quite common. Since the conductance on those 2SK135 2sj50 where in the region of 1A/V so as to drive 7 Amps through them the Vgs had to be Vth+7 V, this could end up being 10 V. Since they where all source followers source ended up hanging 10 V below/above supply rails, rendering the amps bad efficiency. Therefore this arrangement. Let all those power mosfets be where they belong, in switched power supplies. If there is someone interested I have a few pairs of the above mentioned transistors, unused.
Andreas
Andreas
- Status
- Not open for further replies.