I am building a 2 ch single ended dual balanced bridged amp, output transistors are Sanken 2SA1216 and 2SC2922 (200v/17ah/200w) single pair per "bridge" or 4 Sanken's per channel.
This amp would typically run at +-45-55v rails non bridged and I've noticed most bridged amps run on lower rails vs when not bridged.
I am assuming they run at lower rails to compensate for the added gain and higher currents given the supply not being able to scale on currents or in a way the extra power is not needed or risk killing the speakers.
Would you agree this is a correct assumption or am I missing something i.e are there other issues involved keeping the same rail voltages on a bridged amp?
This amp would typically run at +-45-55v rails non bridged and I've noticed most bridged amps run on lower rails vs when not bridged.
I am assuming they run at lower rails to compensate for the added gain and higher currents given the supply not being able to scale on currents or in a way the extra power is not needed or risk killing the speakers.
Would you agree this is a correct assumption or am I missing something i.e are there other issues involved keeping the same rail voltages on a bridged amp?
My guess:
Each amp (or half of each channel in your case) will see half the load impedance, so I would guess that's why the rail voltage would be lower to stay in a safe operating area of the amps/outputs.
Those transistors can stand some abuse I guess, but still giving each amp half of the load impedance may, or may not, be asking a too much, depending on what speakers they will be driving. Bridging two amps with single outputs and driving a nominal 4ohm speaker maybe dipping below 3ohms (1,5 per brigde half) might be too much.
Each amp (or half of each channel in your case) will see half the load impedance, so I would guess that's why the rail voltage would be lower to stay in a safe operating area of the amps/outputs.
Those transistors can stand some abuse I guess, but still giving each amp half of the load impedance may, or may not, be asking a too much, depending on what speakers they will be driving. Bridging two amps with single outputs and driving a nominal 4ohm speaker maybe dipping below 3ohms (1,5 per brigde half) might be too much.
I must be missing something, for the same level with the same load it's the same current but divide by 2 as seen by each transistor pair meaning it's 50% less stressful for the sankens vs non bridged?
Bridges can deliver four times more power (in ideal situation). At both sides of the load the maximum voltage (near rail) but of opposite phase.
Say the rails are +/-50Vdc, at full swing +/-40Vpp at the load. That's 80Vpp on 8Ω (10App). This yields 40Vp/√2 * 5Ap/√2 = 100W (effective sinusodial).With 4Ω one gets (same voltages, twice the current) 40Vp/√2 * 10Ap/√2 = 400W (es).
A non-bridge topo does 20Vp/√2 * 5Ap/√2 = 50W/8Ω and 200W/4Ω (es).
However, for the output bjt's a different calculation is needed, very much depending on the class-setting. In general 120-150% of the output power (at least, per bjt), giving an idea of stress on these chap's. Given the physical heat sink capabilities, lowering the rails is a way to keep the amp alive. Or, in approx 99% of commercial amplifiers, degrading of specs and awful distortion. Bridges are another league, the need massive (dc-) measures.
Say the rails are +/-50Vdc, at full swing +/-40Vpp at the load. That's 80Vpp on 8Ω (10App). This yields 40Vp/√2 * 5Ap/√2 = 100W (effective sinusodial).With 4Ω one gets (same voltages, twice the current) 40Vp/√2 * 10Ap/√2 = 400W (es).
A non-bridge topo does 20Vp/√2 * 5Ap/√2 = 50W/8Ω and 200W/4Ω (es).
However, for the output bjt's a different calculation is needed, very much depending on the class-setting. In general 120-150% of the output power (at least, per bjt), giving an idea of stress on these chap's. Given the physical heat sink capabilities, lowering the rails is a way to keep the amp alive. Or, in approx 99% of commercial amplifiers, degrading of specs and awful distortion. Bridges are another league, the need massive (dc-) measures.
* 2 ch: OK
* single ended : what does that mean?
* dual balanced - bridged: you are being redundant here, in usual acception "bridged" is the most common name and "balanced" a synonym. "Dual" is unnecessary because bridged and balanced NEEDS 2 amplifiers in all cases.
To get same power on same impedance, a bridged amp needs about half supply voltage than a non bridged one.
see above
You must NOT keep same voltages on a bridged amplifier, if both will deliver same power to same speaker impedance.
A very confusing statement and you are NOT referring to maximum power output, just an arbitrary "same voltage" or "same current".
Its a non feedback amp and the input is therefor "dual" or double balanced while the output stage is bridged hence all the words trying to explain even-though it's maybe not the correct way to do it.
I get that and with the same input level you get twice the power which is sort of my point meaning for the same output level one lower the input vs non bridged.To get [B said:
I want more power but don't necessarily need the full double power I would get using the same rail voltage but the extra headroom may be useful provided I don't get other issues.
I am not referring to max power as I will use the amp at similar levels as I do today, just getting a bit more base control, a little less distorsion and better dynamics particularly at higher levels.
Maybe I should rephrase the question, do you think the sankens would be ok on a +-55v rail and 8 ohm load, I assume they would as the amp was happy non bridged with 4 ohm speakers.
The reason I am asking is that I would save on getting new trafo's if I can use the old +-55vdc power-supply.
If the amp has no feedback, output impedance is most likely high. This is bad for bass control as you mention. Bridging will double the output impedance. This will probably do just the opposite of what you are after.. less bass control and dynamics.
Maybe clarify a bit, there is no global feedback but rather local feedback. The amp is in a prototype stage and I am already getting the benefits I described on the test bench so it must be good enough in this respect.
Not sure I follow fully, a non bridged +-40 volt loaded should still peak at 40 volts while the bridged amp can do 80 volts?
It's a class AB running pretty cool at idle but I don't follow the "in general 120-150%"?
Lets say it swing the full railvoltage just to make a point.
40x40/8=200W peak.
80x80/8=800W peak.
40x40/8=200W peak.
80x80/8=800W peak.
And if you got 2ch that swing 40V:
40x40/8=200W peak. if you bridge them, each ch will see half the impedance.
40x40/4=400W peak. x2ch=800W peak.
The real world is not quite like this. You will have som losses in the curcuit. and you will have som PSU sag at high levels.
40x40/8=200W peak. if you bridge them, each ch will see half the impedance.
40x40/4=400W peak. x2ch=800W peak.
The real world is not quite like this. You will have som losses in the curcuit. and you will have som PSU sag at high levels.
I was thinking the same, it's all somewhat irrelevant as the PSU will be the significant limiting factor.
The ones I am running on the test bench have a +-45v 10ah max continuous per channel.
These amps bridged should manage 17ahx2 per channel as long as heat is managed so not sure they would break whatever I do in terms of my current options.
The +-55v supply I was considering has even lower amps, maybe max 5ah per channel continuous.
I would however match that with a significant capacitor bank albeit that would be fused at 6ah on each rail befor the last +- 15,000uF for each channel so all in all maybe this is mostly academic.
I think you have to take a look at SOA in datasheets.
Even if you don't push them to max I guess SOA is limited to about 2 amps at ~70 volts.
Then we got phase upon that. I would have at least doubbled the outputs.
/Figge
Edit: Checked data and SOA is limited to ~2 amps at ~80 volts. Not bad, but still...
Even if you don't push them to max I guess SOA is limited to about 2 amps at ~70 volts.
Then we got phase upon that. I would have at least doubbled the outputs.
/Figge
Edit: Checked data and SOA is limited to ~2 amps at ~80 volts. Not bad, but still...
Last edited:
The SOA curve 2/80 is without heatsink so how do I figure out what it would be with a heatsink and around 40 degrees?
That's the thermal equation you have to do. But first the power handling by the individual bjt's under full load. That's the amount what have to be dissipated. 40°C idle temp only if the die stays well beyond the upper limit, also under full load. And add some safe margin 120-150% - I take 200% to make it robust.
I found the original safe build instructions for the amp and it's as follows:
L C Audio Technology / Millennium X
Trafo/Pwr 8ohm/heatsink/extra sanken pairs
33V AC 128 W 0,5 K/W 0
39V AC 180 W 0,35 K/W 1
24V AC bal. 200 W 0,3 K/W 0
27V AC bal. 300 W 0,2 K/W 0
Again I won't be using it above maybe 120w continuous/average over a few minutes and maybe 250w peak for brief moments.
I will go with a -+40vdc transformer based supply, this supply will "sag" a bit so maybe +-35v when pushed which should keep things sort of safe.
L C Audio Technology / Millennium X
Trafo/Pwr 8ohm/heatsink/extra sanken pairs
33V AC 128 W 0,5 K/W 0
39V AC 180 W 0,35 K/W 1
24V AC bal. 200 W 0,3 K/W 0
27V AC bal. 300 W 0,2 K/W 0
Again I won't be using it above maybe 120w continuous/average over a few minutes and maybe 250w peak for brief moments.
I will go with a -+40vdc transformer based supply, this supply will "sag" a bit so maybe +-35v when pushed which should keep things sort of safe.