Hi, I was wondering would it be safe to run an amp that uses 2sa5200/2sc1943 outputs on 100-0-100 voltage levels?
I have 4 channels , two use original Toshibas with rated max CE voltage of 230v, the other two use Fairchild's analog's with CE rated at 250v.
what do you think?
I have 4 channels , two use original Toshibas with rated max CE voltage of 230v, the other two use Fairchild's analog's with CE rated at 250v.
what do you think?
I tried running mine at 53v with 8ohm speakers and it worked for 3 minutes and then catastrophically failed in shorted out mode and blew fuses on the PSU. So no it won't work. They are cheap to replace though.
I think that if you are considering component choices for a new design, you should know the answer already. The maximum Vceo rating is like a breakdown voltage rating that applies at very small currents. It has little to do with the much lower Vce limit that would apply at maximum dissipation or output power. Look at the SAFE OPERATING AREA graphs for for the max. voltage at say, 2A current using the 100ms line on the datasheets. e.g: 2SC5200 Datasheet pdf - NPN Epitaxial Silicon Transistor - Fairchild Semiconductor
For a single pair of power transistors, maximum power (4 ohm load) is achieved with just 50V rails max. That equates to a Vce of about 90V. Any more voltage and you risk failure. Increasing the number of pairs or the load impedance does allow them to operate at higher voltage/lower current but the law of diminishing returns applies, in my estimation, as you approach 70V supply rails.
The best reason to use these lightweight semis is they are cheap but they are nowhere as powerful as you might think. In $/W terms, MJL4281/4302 would be better and have a much higher SOA limit that will tolerate higher rail voltages. Yes, you do see numbers of naive designs with 2SC5200/A1943 output transistors posted here but we can only assume they are simulated designs that have never been tested at rated maximum power.
For a single pair of power transistors, maximum power (4 ohm load) is achieved with just 50V rails max. That equates to a Vce of about 90V. Any more voltage and you risk failure. Increasing the number of pairs or the load impedance does allow them to operate at higher voltage/lower current but the law of diminishing returns applies, in my estimation, as you approach 70V supply rails.
The best reason to use these lightweight semis is they are cheap but they are nowhere as powerful as you might think. In $/W terms, MJL4281/4302 would be better and have a much higher SOA limit that will tolerate higher rail voltages. Yes, you do see numbers of naive designs with 2SC5200/A1943 output transistors posted here but we can only assume they are simulated designs that have never been tested at rated maximum power.
well I have 4 similar channels , each channel has 4 pairs of these transistors , previously I ran them on 90-0-90 and they were fine,
I'm just thinking because I will have to check how much voltage I'm getting from my self made smps transformer , I was just a bit lazy and thought that maybe if it's not over +-200 it should be ok.
also what do you think with 4 pairs whats the lowest impedance I could run , could it go for 2 ohms? probably not 4 would be safe i assume.
I'm just thinking because I will have to check how much voltage I'm getting from my self made smps transformer , I was just a bit lazy and thought that maybe if it's not over +-200 it should be ok.
also what do you think with 4 pairs whats the lowest impedance I could run , could it go for 2 ohms? probably not 4 would be safe i assume.
Did you test the amplifiers you built at full rated power with 90V rails or just use them at low power and assume they would survive at max. power? Maximum power means full output voltage at the onset of clipping with a standard signal like 1kHz sine wave. The load needs to specified but should be resistive only at 4, 8 ohms etc.
Few constructors want to know the weakness in their designs and find load testing too difficult or too cruel on their project. If you don't test though, you can't claim an amplifier "works" at any particular supply voltage and power level.
Few constructors want to know the weakness in their designs and find load testing too difficult or too cruel on their project. If you don't test though, you can't claim an amplifier "works" at any particular supply voltage and power level.
It would take 6 pairs to do 8 ohms, 12 to do 4. If you are making your own SMPS transformer, consider running H-class with 100 and 50V taps. The taps dont cost anything - just an extra set of rectifiers and caps, and you have to wind it anyway.Then you can use 4 pairs and its perfectly reliable at 4 ohms.
2SC5200/2SA1943
Secondary breakdown is a serious problem at these higher rails.
If you want to drive a 4 ohm speaker with +/-100v rails you will require 7+7 devices at 25 deg C
At 70 deg C (A better design with SOA temperature derating) you require 10+10 devices.
Now if you run your devices in series, secondary breakdown is pretty much eliminated as you are effectively running each series stack at +/- 50v, you now need 2+2 devices per stack (A total of 8 devices as compared to 20 and little chance of secondary breakdown.
Yes you need extra drivers and predrivers for the series stack but in total you have a better design.
We have been using these devices for over 30 years (We of course used the older 2SC3281/2SA1302 in those years) and never ran them over a +/-55v rail per stack. In lower powered amplifiers we did not stack them of course.
ALL bipolars used for audio benefit from lower supplies and higher current
Steve Mantz
Zed Audio Corp.
Secondary breakdown is a serious problem at these higher rails.
If you want to drive a 4 ohm speaker with +/-100v rails you will require 7+7 devices at 25 deg C
At 70 deg C (A better design with SOA temperature derating) you require 10+10 devices.
Now if you run your devices in series, secondary breakdown is pretty much eliminated as you are effectively running each series stack at +/- 50v, you now need 2+2 devices per stack (A total of 8 devices as compared to 20 and little chance of secondary breakdown.
Yes you need extra drivers and predrivers for the series stack but in total you have a better design.
We have been using these devices for over 30 years (We of course used the older 2SC3281/2SA1302 in those years) and never ran them over a +/-55v rail per stack. In lower powered amplifiers we did not stack them of course.
ALL bipolars used for audio benefit from lower supplies and higher current
Steve Mantz
Zed Audio Corp.
Well I'm not a fan of class H , these were actually class H first but they also had the lower power version if you take the mosfet drivers out it turns into a class AB design.It's the apex h900/500 version I'm talking about right here from the forum , I just made it some years back and now finally decided to put it in a box.
haven't load tested yet.
So if I would instead used the more rugged MJL4281/4302 pairs I could then probably do 100v rails with no problem.
haven't load tested yet.
So if I would instead used the more rugged MJL4281/4302 pairs I could then probably do 100v rails with no problem.
The transistor count on those boards was intended for class H. If you're going to run class AB use as rugged a device as you can. That's either the 4281 or 21196. Converting it to a Super-Leach output stage is another possibility. You could build little add-on boards for the slave transistor banks and their associated bias resistors. The slave transistors do not need thermal bias compensation so they can even be heat sinked separately. Just keep the wiring short and sweet or you'll make an oscillator.
I used to run a pair of homebrew monoblocks that used 10 pair of 21193/4 in series/parallel from +/-127V. Sub-2 ohm load at high output (pair of labhorns per amp) all day even out in hot sun. The last time I used them I ended up turning the plugs on the extension cords into molten goo so they got mothballed and replaced with more efficient amps.
I used to run a pair of homebrew monoblocks that used 10 pair of 21193/4 in series/parallel from +/-127V. Sub-2 ohm load at high output (pair of labhorns per amp) all day even out in hot sun. The last time I used them I ended up turning the plugs on the extension cords into molten goo so they got mothballed and replaced with more efficient amps.
well , sadly there isn't much space in the box anymore , but I already downgraded the secondary of the transformer to 90-0-90 volts idle so in reality about 85v under load.
also the amp has bridge mode and since i have 4 separate channels i can still have stereo even using bridge that makes each channel have 8 pairs of transistors so I hope i will suffice with that if I need to drive lower impedance higher power loads.
what do you think?
also the amp has bridge mode and since i have 4 separate channels i can still have stereo even using bridge that makes each channel have 8 pairs of transistors so I hope i will suffice with that if I need to drive lower impedance higher power loads.
what do you think?
Bridging is not what you want to do driving low impedance loads. If anything, you want to find a way to parallel channels. Maybe fully populate only two boards, and the output transistors only on the other two. Then run the extra set of outputs in parallel with the first. One of the big Adcom amps does this.
the SOA of the 1943/5200 is near the worst of any power device.
They do not have much current capability at higher voltages.
datasheet shows 3A @ 45Vce, but this drops to 1A @ 70Vce and 0.38A @ 100Vce
That is just 38W from a 150W device due to poor SOA @ high voltage.
compare to a 2n3773 which is guaranteed by 100% testing to do 150W @ 100Vce and it's also a 150W device rated at only 140Vce0
CrazyM says
They do not have much current capability at higher voltages.
datasheet shows 3A @ 45Vce, but this drops to 1A @ 70Vce and 0.38A @ 100Vce
That is just 38W from a 150W device due to poor SOA @ high voltage.
compare to a 2n3773 which is guaranteed by 100% testing to do 150W @ 100Vce and it's also a 150W device rated at only 140Vce0
CrazyM says
Last edited:
It's not "the worst" by a long shot. It's "good enough" for 50 volt supplies and 8 ohm loads. Which is the majority of store-bought home audio gear. You're not really supposed to be building 200+ watt amps with them unless you operate in some class that reduces VCE. Many older types and most switching types have second breakdown points of 20 volts or less. And they fall off in power capacity MUCH faster versus VCE than Toshiba audio types. Those types wouldn't work even in an old NAD 3020.
well i've reduced the voltage to about 85-0-85 and quite frankly since I will ahve protection also etc I might just test them to what they ca and see what happens , overall it's not that bad the amp itself is a sort of experiment more due to the psu being smps my first high power smps so I got these left over and decided to add them as load.
much bigger of an issue is the parasitic ripple i'm having at my smps output that I simply can't get rid of.
bought the etf59 core with an N87 grade ferrite , put snubbers across secondary cap bank added big enough capacitance yet still.
maybe cause i'm using IGBT's for switching due to them being more robust than mosfets and much cheaper for same current capability.
and still i hate semiconductors at high voltage/current applications such as high power smps, the slightest mistake and puff goes your money and time.faster than you can probe with a scope to see the problem etc.
well with all it's noise atleast i have managed to make t last large power loads , attached some old irons each 1600w mains rated and a halogen bulb and a water heater and only dropped some 10 to 15 volts being unregulated quite nice.and that was before while using a God knows what type of ferrite core that i got who knows where.
haven't tested with the N87 but will soon.
yet again every slight thing comes with it's surprises this transformer now is getting bit warm at idle I wonder why.I'm having 11 or 12 turns primary can't remember with a really thick litz and 6-6 turns secondary being a half bridge the primary runs half the mains rectified which for me is 325/2
much bigger of an issue is the parasitic ripple i'm having at my smps output that I simply can't get rid of.
bought the etf59 core with an N87 grade ferrite , put snubbers across secondary cap bank added big enough capacitance yet still.
maybe cause i'm using IGBT's for switching due to them being more robust than mosfets and much cheaper for same current capability.
and still i hate semiconductors at high voltage/current applications such as high power smps, the slightest mistake and puff goes your money and time.faster than you can probe with a scope to see the problem etc.
well with all it's noise atleast i have managed to make t last large power loads , attached some old irons each 1600w mains rated and a halogen bulb and a water heater and only dropped some 10 to 15 volts being unregulated quite nice.and that was before while using a God knows what type of ferrite core that i got who knows where.
haven't tested with the N87 but will soon.
yet again every slight thing comes with it's surprises this transformer now is getting bit warm at idle I wonder why.I'm having 11 or 12 turns primary can't remember with a really thick litz and 6-6 turns secondary being a half bridge the primary runs half the mains rectified which for me is 325/2
Last edited:
- Status
- Not open for further replies.