We are well into the New year 
so its good we are getting back into discussing amplifiers again, more specifically , i would like to discuss were we left off on Building amps for low-z load .
Question:
I would love to discuss different methodology on lowering distortion when driving low-z loads, more Bias? More Fb, increase output stage....?
so its good we are getting back into discussing amplifiers again, more specifically , i would like to discuss were we left off on Building amps for low-z load . Question:
I would love to discuss different methodology on lowering distortion when driving low-z loads, more Bias? More Fb, increase output stage....?
If you do not use class A bias, then idle current should be 'optimal', i.e. 25mV accross Re. For Re = 0R22 it makes 114mA per pair and in case of 6 pairs one has 0.68A idle current .... pretty high and quite a lot of heat.
NFB reduces THD, but should not be used as the only cure and of course one has to watch stability.
NFB reduces THD, but should not be used as the only cure and of course one has to watch stability.
That's almost exactly what I do, however I use "only" 4 pairs of output transistors, so I raise the bias just a bit, usually not over 130 mA per trannie. I believe if one errs, then a bit more is better than a bit less.
So I end up with 520 mA as my total bias. Yes, the heatsinks do get warm, but then, I want them to bexause if they do so on bias, then most of my program material will not either influence or be influenced by thermal considerations under usual circumstances.
Wayne, Pavel just told you what you should do. Tremendous current requirements need a good number of output devices to share the load and still have reserve left. How many exactly depends also on you PSU line voltages, the higher they are, the still more trannies you will need.
Some number of messages ago, Abraxalito told you the whole story - go for relatively low PSU line voltages, +/- 42V, as this gives you reasonable power into higher Z loads, but is low enough to help you get away with a still reasonable number of power devices. Condiser - according to its data sheet, a Motorola/ON Semi MJL 3281 A transistor is capable of delivering 4.5 Amps at 25 deg C. However, from this you need to derate power as per output device heating up, and given that you will heat them up a lot since you will need high current levels, the real world deliverable power will depend on heating and cooling a lot. This means you will need mighty heat sinks. Assuming heat sink temp is limited to 65 deg., the inside of the trannies isat least 20 deg. higher, so your nominal dissipation drops from 200W per device to some 114W. If your PSU line 42V, and you lose those 2V on drops across predriver/driver/power transistors, that leaves you just (114:40) 2.85A.
So, obviously if you want to have 20A on call, you will need 7 pairs of those devices, round it off to 8 pairs to be safe.
Obviously, this is just a rough calculation, but it sure is indicative.
I would hate to be a heat sink in your amp.
Yes D, that part is pretty academic ( output stage soa ) i think i need to rephrase my question .
All amplifiers i have seen have more distortion at 2 ohm vs 8 ohm , how do we reduce the increase in distortion when driving low-z loads , getting them closer or to match its 8 ohm numbers .
Is this even possible ..?
PS: when i do build the DVV special it will be running on 42v ...
All amplifiers i have seen have more distortion at 2 ohm vs 8 ohm , how do we reduce the increase in distortion when driving low-z loads , getting them closer or to match its 8 ohm numbers .
Is this even possible ..?
PS: when i do build the DVV special it will be running on 42v ...
Yes D, that part is pretty academic ( output stage soa ) i think i need to rephrase my question .
All amplifiers i have seen have more distortion at 2 ohm vs 8 ohm , how do we reduce the increase in distortion when driving low-z loads , getting them closer or to match its 8 ohm numbers .
Is this even possible ..?
PS: when i do build the DVV special it will be running on 42v ...
Theoretically, ANYTHING is possible.
It's a simple fact that transistors distort more when dissipating more current, hence 4 Ohm distortion figures are always worse than 8 Ohm ones. The lower down you go, the harder it becomes to control distortion. Basically, there are two ways to go.
One is to add more feedback. In Frank's case, that means something like 86 dB (20,000:1), in my case that's more like 40 dB (100:1).
The other is to add still more output devices. Since they are sharing the load, this will offload individual transistors and thus reduce distortion. Given the specific job at hand, I'd switch over to TO-3 metal cased MJ2xxxx series, for several reasons - they were designed for audio and have very low distortion, they are individualy rated at higher power levels and thus deal with heating better than plastic case trannies, etc.
Probably a combination of these two methods will fare best.
Tell you what, I'll sim a bit tomorrow, just to get the feel of it.
Oh yeah, you still owe me a family pack of the hot sauce you said was way better than Tabasco.
a.wayne, looks like we're on the same road for a bit ... I'm having an interesting tussle at the moment, seeing if I can achieve a realistic amplifier construction, in sim world, that can do 800W into 1 ohm at -120dB distortion. Stability will be the elephant in the room, the unit needs to be able to do this without disintegrating into headless chicken oscillation at the slightest provocation - a nice New Year challenge ...
If the power amp has zero output impedance.
You don't like MOSFET output stages, Mr Wayne ?
Output transformers.
Not only to get lower distortion from the output stage but also to reduce power supply noise.Providing more and more output pairs increases the capacitive (via Ccb) coupling of the PSU to the signal. PSRR is going to degrade with each additional output pair.
Not adverse to anything but bad sound Dutchie, no preference at all .....
Output transformers.
Providing more and more output pairs increases the capacitive (via Ccb) coupling of the PSU to the signal. PSRR is going to degrade with each additional output pair.
Yes , but neither is passing the signal thru miles of copper and transformer hysteresis an answer ...
You first , tell me about your SS amp with transformer outputs and while at it , name one commercial amp from 1 dollar to 1 mill which uses one in the output , well apart from Mac ..
It does appear most are willing to live with the higher capacitance and lower Psrr ,
Just saying ...
It does appear most are willing to live with the higher capacitance and lower Psrr ,
Just saying ...
Last edited:
- Status
- Not open for further replies.