Have the output stage wired up and all running off my max +/- volt bench supply. Bias for now is set for 100mv across the emitter resistors and the sink has definately gone up in temperature a bit but at this voltage level is just warm to the touch. I highly reccomend testing it all this way before applying the full tilt of your supply.
Does anyone know offhand how many mv across the emitter resistors shold be measured at +/-45 volts? The KSA 80 service info says 325 mv. WOuld this also be correct for thi KSA-50? I've used .47 ohm emitter resistirs in my amp....
Mark
Does anyone know offhand how many mv across the emitter resistors shold be measured at +/-45 volts? The KSA 80 service info says 325 mv. WOuld this also be correct for thi KSA-50? I've used .47 ohm emitter resistirs in my amp....
Mark
With the dc offset - it makes no difference whatsoever at the moment...
i get about 0.4v across the emitter resistors and about 0.8v across the 2 output connections, which leads me to believe that there may be something up...
I used 47pf caps in mine, not sure if it's oscillating, as i don't have a scope to check with, however i do have a fluke multimeter that does do hz, so i might check later to see if it is getting any freq's on the drivers...
I still can't work out why i am getting so much voltage... Possibly the markings on my toroid are wrong and i am getting more like 32v rather than the 28 it says it does, but it is unloaded with very large caps on it...Possibly i should use a fuse too.... (shhh don't tell anyone!)
I measure everything in deg C...
Aaron
P.S. Mark - what do you get in measurements across the n output and p output of the board itself?
P.P.S. - that bias sounds about right to me, about 0.6amp, you may be able to back that off a little bit possibly, for testing
i get about 0.4v across the emitter resistors and about 0.8v across the 2 output connections, which leads me to believe that there may be something up...
I used 47pf caps in mine, not sure if it's oscillating, as i don't have a scope to check with, however i do have a fluke multimeter that does do hz, so i might check later to see if it is getting any freq's on the drivers...
I still can't work out why i am getting so much voltage... Possibly the markings on my toroid are wrong and i am getting more like 32v rather than the 28 it says it does, but it is unloaded with very large caps on it...Possibly i should use a fuse too.... (shhh don't tell anyone!)
I measure everything in deg C...
Aaron
P.S. Mark - what do you get in measurements across the n output and p output of the board itself?
P.P.S. - that bias sounds about right to me, about 0.6amp, you may be able to back that off a little bit possibly, for testing
Across each drivers emmiter resistor I am getting .757 and .759 volts respectively. I think that for the DC balance on mine to work I need higher rail voltages in order for the zeners to start conducting. Then there is something to refrence the zero adjust to.... I shall see tonight or tommrrow depending on the time frame. I do have my toroids and caps... just need to wire them up and gradually crank up the variac and see what happens.
Mark
Mark
Running +/- 43 volts on the rails I am almost at 2.4 amps bias and the output semis are a tad too hot at 76 deg c. after an hour! So more sinks or fans is needed in this case. Also a DC balance problem (200 mv) with this board for what ever reason. Will double check the pc fix jumpers to be sure thay are correct....... All else seems ok and it sounds pretty good too..... I will split up my output devices between two heat sinks. No signs of any oscillation at all. BTW: I am using those poisonous ceramic/alumnium oxide insulators on this thing to help up the efficiency of heat transfer any way I can......
http://www.aavidthermalloy.com/products/standard/access/insulate.shtml
Mark
http://www.aavidthermalloy.com/products/standard/access/insulate.shtml
Mark
So i got my test mule up and running!
+-43vdc i got (from a 28vac toroid...so i may have to look at a lower voltage one!) from my p/s, i ran 1 set of outputs @ approx 0.1v bias - which gives me about 200ma or so bias i think (??). Because they were on a 2.7c/w heatsink, and there was only one set of outputs, this made the heatsinks put out about 60deg heat (total)....After the boards "power down" - ie after power is d/c they do do odd funny sounds through the speaker (even with no input) and as it gets to the lower region (about 20v power supply) puts dc across the speaker... Unless this is just mine - solution is to simply use a relay to d/c the speaker outputs once power is off! I had it running for a fair while (about 30mins) and as the heatsinks got slightly slightly hotter (1deg C) the bias was still going up by 0.001v every 5mins or so, so it was getting into thermal runaway, just very very slowly... possibly not helped by tiny heatsink, a lot of bias and high rails...
What does everyone think? Sound right? I listened to it - other than mains interference (wiring everywhere and using fig 8 wire as "rca's"... probably didn't help!)... I used 47pf caps and i haven't got a scope to test unfort...
What does everyone think about the amp going into DC on the speaker out's? Is that normal? (be aware i am using centre tap (0v) as the speaker -ve and it doesn't have any form of star ground (not that far into wiring it!) - would that be the problem?
the pre drivers will need at least a small heatsink, they do get hot, stable about 70deg C tonight (it's about 25 ambient again)....
the "mains" interference i was getting was coming and going too - i.e it would come on for a while then dissapear - say when the volume was turned up or the channel was changed (tv channels).... I figure it needs a star ground and also needs decent wiring - not wires running everwhere...
My dc offset is around -30mv... every time i touch my pot with a screwdriver it goes crazy, makes it hard to set - bias however is easy to set!
Aaron
+-43vdc i got (from a 28vac toroid...so i may have to look at a lower voltage one!) from my p/s, i ran 1 set of outputs @ approx 0.1v bias - which gives me about 200ma or so bias i think (??). Because they were on a 2.7c/w heatsink, and there was only one set of outputs, this made the heatsinks put out about 60deg heat (total)....After the boards "power down" - ie after power is d/c they do do odd funny sounds through the speaker (even with no input) and as it gets to the lower region (about 20v power supply) puts dc across the speaker... Unless this is just mine - solution is to simply use a relay to d/c the speaker outputs once power is off! I had it running for a fair while (about 30mins) and as the heatsinks got slightly slightly hotter (1deg C) the bias was still going up by 0.001v every 5mins or so, so it was getting into thermal runaway, just very very slowly... possibly not helped by tiny heatsink, a lot of bias and high rails...
What does everyone think? Sound right? I listened to it - other than mains interference (wiring everywhere and using fig 8 wire as "rca's"... probably didn't help!)... I used 47pf caps and i haven't got a scope to test unfort...
What does everyone think about the amp going into DC on the speaker out's? Is that normal? (be aware i am using centre tap (0v) as the speaker -ve and it doesn't have any form of star ground (not that far into wiring it!) - would that be the problem?
the pre drivers will need at least a small heatsink, they do get hot, stable about 70deg C tonight (it's about 25 ambient again)....
the "mains" interference i was getting was coming and going too - i.e it would come on for a while then dissapear - say when the volume was turned up or the channel was changed (tv channels).... I figure it needs a star ground and also needs decent wiring - not wires running everwhere...
My dc offset is around -30mv... every time i touch my pot with a screwdriver it goes crazy, makes it hard to set
- bias however is easy to set!Aaron
it's not unusual...
...to get short periods of weirdness as amps power on and off. Once you have the bias set to the normal value the supply will collapse much faster (20x) and whatever happens will be much shorter in duration.
You may find the offset easier to deal with once the rest of the outputs are in place and extra current is making the dampening effect of the emitter resistors more pronounced. At this point, a change of .05v across an emitter resistor is doubling or halving your standing current...
Almost makes me wish I was back at home to mess with my own...
Stuart
...to get short periods of weirdness as amps power on and off. Once you have the bias set to the normal value the supply will collapse much faster (20x) and whatever happens will be much shorter in duration.
You may find the offset easier to deal with once the rest of the outputs are in place and extra current is making the dampening effect of the emitter resistors more pronounced. At this point, a change of .05v across an emitter resistor is doubling or halving your standing current...
Almost makes me wish I was back at home to mess with my own...
Stuart
"What does everyone think about the amp going into DC on the speaker out's?"
Well, some amps will do this, Phase Linears let a small burp a few seconds after shut down. So far my KSA-50 doesn't have a hint of that problem on power up or down. On power down it runs for about 5 seconds then settles into distortion for about another 5 seconds like most amps will do, but no DC on the output. Am using 68,000 mfd per rail right now....
But man this thing is a real scorcher
with my sinks settling in at 76 deg c. which is way too hot, and half way to the max rating of 150 c. for these plastic transistors. I am running .47 ohm emitter resistors and may go to .68 ohm and see what happens..... The bias doesn't seem to creep after about 45 min or so, settles in nicely but I only got to 300mv across the emitter resistors. So I either have to set up a wind tunnel with these sinks or at least double the amount of sink that I am using.
Mark
Well, some amps will do this, Phase Linears let a small burp a few seconds after shut down. So far my KSA-50 doesn't have a hint of that problem on power up or down. On power down it runs for about 5 seconds then settles into distortion for about another 5 seconds like most amps will do, but no DC on the output. Am using 68,000 mfd per rail right now....
But man this thing is a real scorcher
with my sinks settling in at 76 deg c. which is way too hot, and half way to the max rating of 150 c. for these plastic transistors. I am running .47 ohm emitter resistors and may go to .68 ohm and see what happens..... The bias doesn't seem to creep after about 45 min or so, settles in nicely but I only got to 300mv across the emitter resistors. So I either have to set up a wind tunnel with these sinks or at least double the amount of sink that I am using.Mark
heatsinks
For my alephs I use a 4 squarish sinks to make 4 sides of a cube, ~10" on a side, and can get ~150w out of the thing at ~50c, each fin is ~1.25 inch, and I think there are 4 fins to an inch on a 1/4" base.
Mark, with 43 volt rails and 2.4amps, you have a standing dissipation of 200+watts per channel, so you will need to scale accordingly. I'll be using one of my cubes for each channel, rails at ~35, and I'll be cranking the bias until I get to ~55c on the middle of each face. I use a digital kitchen thermometer, so I'm not sure how accurate it is, but using the touch test I'm in the ballpark.
I'm not sure I can keep a pair of krell amps in cuboid form, it's kind of like dropping a ford engine in a chevy. I think cubes are reserved for alephs, but it a very efficient way of making a chassis, I'll have to figure something else out...
Stuart
For my alephs I use a 4 squarish sinks to make 4 sides of a cube, ~10" on a side, and can get ~150w out of the thing at ~50c, each fin is ~1.25 inch, and I think there are 4 fins to an inch on a 1/4" base.
Mark, with 43 volt rails and 2.4amps, you have a standing dissipation of 200+watts per channel, so you will need to scale accordingly. I'll be using one of my cubes for each channel, rails at ~35, and I'll be cranking the bias until I get to ~55c on the middle of each face. I use a digital kitchen thermometer, so I'm not sure how accurate it is, but using the touch test I'm in the ballpark.
I'm not sure I can keep a pair of krell amps in cuboid form, it's kind of like dropping a ford engine in a chevy. I think cubes are reserved for alephs, but it a very efficient way of making a chassis, I'll have to figure something else out...
Stuart
derating matters here...
Running the transistors very hot will lower your class B reserve, you lose 1.43w of capacity for each degree C above 25c. At 75c, the junction is at 105c or more, and your 200w studly BJT is down to 85w and already using 30+ of that capacity...fans are your friends...
Ignore the max. temp rating, it never had any meaning and it will mislead you into thinking your amp could survive really high temps, it won't. By the time the heatsink is at 100c+, the transistor will be on it's last legs...Try redrawing the SOA curve with the 'new' dissipation, you'll see at this temp the transistor is down to the point where the idle dissipation alone is all it can manage, virtually no margin or class B at all.
Stuart
P.S. If you need 0.68R resistors I got you covered, drop me a line. Not sure if they are inductive, pinkmouse thinks they might be stacked film, I have no way of knowing, if anyone else has an idea, especially those that have received theirs, please let us know...
Running the transistors very hot will lower your class B reserve, you lose 1.43w of capacity for each degree C above 25c. At 75c, the junction is at 105c or more, and your 200w studly BJT is down to 85w and already using 30+ of that capacity...fans are your friends...
Ignore the max. temp rating, it never had any meaning and it will mislead you into thinking your amp could survive really high temps, it won't. By the time the heatsink is at 100c+, the transistor will be on it's last legs...Try redrawing the SOA curve with the 'new' dissipation, you'll see at this temp the transistor is down to the point where the idle dissipation alone is all it can manage, virtually no margin or class B at all.
Stuart
P.S. If you need 0.68R resistors I got you covered, drop me a line. Not sure if they are inductive, pinkmouse thinks they might be stacked film, I have no way of knowing, if anyone else has an idea, especially those that have received theirs, please let us know...
O found my DC balance problem... had just jumpered a wrong trace. Nuttr may want to double check his jumpers too. My offset is now 0 volts and the voltages across the drivers emitter resistiors are nor balanced as a result. I've also increased my output transistor's emitter resistors to .68 ohms and am now back up at 100 mv across them. Am goping to slowly crank it up over a couple of hours again and see how it goes.
CUBE! Thats definately reserved ONLY for Alephs...... Man you'd have a small 400 watt cube sitthing there..... Either those big huge square voltage regulator finned heat sinks or air tunnel arrangement wuill be accepted! I think Dan originally found out forced air was the only way. I am using fairly large sinks with 2" fins and with two of them facing each other with two 3" fans running very slowly ... just fast enough to keep the sinks at 50 deg c..... That should work and the dual fans will also be redundant. I will still use a clixon sensor on each heat sink just to be safe. What I may do is to have the clixon flip the amp into a lower bias mode instread of turning it off......
Mark
CUBE! Thats definately reserved ONLY for Alephs...... Man you'd have a small 400 watt cube sitthing there..... Either those big huge square voltage regulator finned heat sinks or air tunnel arrangement wuill be accepted! I think Dan originally found out forced air was the only way. I am using fairly large sinks with 2" fins and with two of them facing each other with two 3" fans running very slowly ... just fast enough to keep the sinks at 50 deg c..... That should work and the dual fans will also be redundant. I will still use a clixon sensor on each heat sink just to be safe. What I may do is to have the clixon flip the amp into a lower bias mode instread of turning it off......
Mark
OK, I got a pair of 4" fans running on this sink and have it cranked to 2.4 amps as measured on the negative rail. I hope this is a per rail spec and not total........ This amounts to about 520 mv across what are now .68 ohm emitter resistors(down form .47 ohm) as in the original KSA-50, and about 1.1 volts across each of the driver resistors. Its interesting to note that ALL edginess has disappeared as well. I can guarantee you that without these fans the fire department would been here and gone! BTW: Device temp is at just shy of 60 deg c. as measured on the center of the three cases with my Tektronix temperature probe..... Still seems a bit on the high side to me and this is with both fans running at about half rated voltage. Water cooling might even be a better solution (literally!).
Your above post has me a little confused about loosing class b headroom????? were not the original KSA-50's biased this high? I am sassuming they were and thet biasing any lower would basically keave me with a KST-100 (or thereabouts). I don't remember my KSA-80B ever running nearly as hot as this channel of KSA-50.....
Mark
This amounts to about 520 mv across what are now .68 ohm emitter resistors(down form .47 ohm) as in the original KSA-50, and about 1.1 volts across each of the driver resistors. Its interesting to note that ALL edginess has disappeared as well. I can guarantee you that without these fans the fire department would been here and gone! BTW: Device temp is at just shy of 60 deg c. as measured on the center of the three cases with my Tektronix temperature probe..... Still seems a bit on the high side to me and this is with both fans running at about half rated voltage. Water cooling might even be a better solution (literally!).Your above post has me a little confused about loosing class b headroom????? were not the original KSA-50's biased this high? I am sassuming they were and thet biasing any lower would basically keave me with a KST-100 (or thereabouts). I don't remember my KSA-80B ever running nearly as hot as this channel of KSA-50.....
Mark
depends...
...at the point where you run out of class A current, at approx 2 * idle, one pole of the output is off, so no problem with power there. But the other is now being used to do more current than normal, so depending on the nature of the load it is going to be asked to dissipate more power, somewhere between a little, a lot and too much.
If you are running close to the ragged edge, there is always the chance the magic smoke will be let out, and in the case of an amp with power supplies like this one, there will be a lot of magic smoke...
Keep 'em cool and you increase the margin your transistors have for those special welding moments...
IIRC, the KSA50 had 36v rails and 1.8amps bias, while the KSA100 had 47v rails and 2.7amps. Lloyd Maclean (?sp) gave the details he was aware of. So with 43v/2.4amp you are splitting the difference. The ksa50 had 2 pairs of to3's, while the ksa100 had 4 pairs of to3's. Not sure how many of the plastic guys you have, but scale accordingly. IMHO if you are planning on hot running, twice as many, if you can keep them cool, 50% more.
Glad to hear everything else is going as planned,
Stuart
...at the point where you run out of class A current, at approx 2 * idle, one pole of the output is off, so no problem with power there. But the other is now being used to do more current than normal, so depending on the nature of the load it is going to be asked to dissipate more power, somewhere between a little, a lot and too much.
If you are running close to the ragged edge, there is always the chance the magic smoke will be let out, and in the case of an amp with power supplies like this one, there will be a lot of magic smoke...
Keep 'em cool and you increase the margin your transistors have for those special welding moments...
IIRC, the KSA50 had 36v rails and 1.8amps bias, while the KSA100 had 47v rails and 2.7amps. Lloyd Maclean (?sp) gave the details he was aware of. So with 43v/2.4amp you are splitting the difference. The ksa50 had 2 pairs of to3's, while the ksa100 had 4 pairs of to3's. Not sure how many of the plastic guys you have, but scale accordingly. IMHO if you are planning on hot running, twice as many, if you can keep them cool, 50% more.
Glad to hear everything else is going as planned,
Stuart
They were surprisingly not very warm. But then there was a fan blowing through a mid sized wind tunnel. HS temp was lower than 60C. The HS was as wide as any wind tunnel and about 4-5 inches long.
If my calcs were right, here are the stats:
Class-A output Calculation: Krell KSA-100 Mk. II
Inputs
Number of output devices: 4 Pairs
Voltage rails (per rail): 53 volts
Emitter resistnce (per device): 1 ohms
voltage per Emmiter resistor: 690 mv
Idle bias per device 690.0 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.690 Amps
Total Amplifier bias (per rail) 2.76 Amps
Total Amplifier bias (both rails) 5.52 Amps
Dissipation (per channel) 292.6 Watts
Dissipation per device pair at idle 36.6 watts
Class-A output: Peak 243.8 Watts peak
Class-A output: RMS 121.9 Watts RMS
Efficiency 41.66 %
If my calcs were right, here are the stats:
Class-A output Calculation: Krell KSA-100 Mk. II
Inputs
Number of output devices: 4 Pairs
Voltage rails (per rail): 53 volts
Emitter resistnce (per device): 1 ohms
voltage per Emmiter resistor: 690 mv
Idle bias per device 690.0 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.690 Amps
Total Amplifier bias (per rail) 2.76 Amps
Total Amplifier bias (both rails) 5.52 Amps
Dissipation (per channel) 292.6 Watts
Dissipation per device pair at idle 36.6 watts
Class-A output: Peak 243.8 Watts peak
Class-A output: RMS 121.9 Watts RMS
Efficiency 41.66 %
Calculations for KSA-50
Here calcs dome for our KSA-50 clone:
Class-A output Calculation: Krell KSA-50 Mk. II
Inputs
Number of output devices: 3 Pairs
Voltage rails (per rail): 36 volts
Emitter resistance (per device): 0.68 ohms
Bias voltage per Emmiter resistor: 400 mv
Idle bias per device 588.2 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.588 Amps
Total Amplifier bias (per rail)** 1.76 Amps
Total Amplifier bias (both rails) 3.53 Amps
Total Dissipation (per channel) 127.1 Watts
Dissipation per device pair at idle 21.2 watts
Class-A output: Peak 99.7 Watts peak
Class-A output: RMS 49.8 Watts RMS
Efficiency 39.22 %
Here calcs dome for our KSA-50 clone:
Class-A output Calculation: Krell KSA-50 Mk. II
Inputs
Number of output devices: 3 Pairs
Voltage rails (per rail): 36 volts
Emitter resistance (per device): 0.68 ohms
Bias voltage per Emmiter resistor: 400 mv
Idle bias per device 588.2 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.588 Amps
Total Amplifier bias (per rail)** 1.76 Amps
Total Amplifier bias (both rails) 3.53 Amps
Total Dissipation (per channel) 127.1 Watts
Dissipation per device pair at idle 21.2 watts
Class-A output: Peak 99.7 Watts peak
Class-A output: RMS 49.8 Watts RMS
Efficiency 39.22 %
For Mark, this is what he did:
Class-A output Calculation: Krell KSA-50 Mk. II
Inputs
Number of output devices: 3 Pairs
Voltage rails (per rail): 36 volts
Emitter resistance (per device): 0.68 ohms
Bias voltage per Emmiter resistor: 520 mv
Idle bias per device 764.7 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.765 Amps
Total Amplifier bias (per rail)** 2.29 Amps
Total Amplifier bias (both rails) 4.59 Amps
Total Dissipation (per channel) 165.2 Watts
Dissipation per device pair at idle 27.5 watts
Class-A output: Peak 168.4 Watts peak
Class-A output: RMS 84.2 Watts RMS
Efficiency 50.98 %
These calcs do NOT take into consideration any losses. Looks like Mark is "cooking" his OP devices.
Class-A output Calculation: Krell KSA-50 Mk. II
Inputs
Number of output devices: 3 Pairs
Voltage rails (per rail): 36 volts
Emitter resistance (per device): 0.68 ohms
Bias voltage per Emmiter resistor: 520 mv
Idle bias per device 764.7 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.765 Amps
Total Amplifier bias (per rail)** 2.29 Amps
Total Amplifier bias (both rails) 4.59 Amps
Total Dissipation (per channel) 165.2 Watts
Dissipation per device pair at idle 27.5 watts
Class-A output: Peak 168.4 Watts peak
Class-A output: RMS 84.2 Watts RMS
Efficiency 50.98 %
These calcs do NOT take into consideration any losses. Looks like Mark is "cooking" his OP devices.
