No, bootstrap gives max voltage, max current.
Collectors go to the bootstrap as well so in high voltage applications.
Your Q3 and Q4 could be lower voltage transistors with the bootstrap.
A set resistor doesn't maintain constant current, especially a low one like 470 ohms.
With constant voltage from a regulated supply it somewhat works.
Regardless 470 ohms is a ridiculous value. Your in sim but dont realize
your blasting Q3 and Q4 with 30ma of current? when they dont need more than 3 to 10ma.
If your peak to peak is 13.2 volts be a little more than 9 volts RMS on the output.
Your not going to put 10 watts into 8 ohms with TO-126 transistors.
The gain will fall flat on its face and the actual realistic current you could put into 8 ohms
was probably what voltage swing you'll get without the capacitor.
Does LTspice have real time scope? must not.
You must be showing a very very short transient. To get 13.2 peak to peak.
That would last in real life for about 2 seconds. till that cap charge falls flat on its face.
After 2 seconds the voltage swing be the same as without the cap, but with huge crossover distortion.
Your circuit will do full swing into 600 ohms without the cap maybe 50 ohms at most.
Then the pointless cap can be removed, and with the bootstrap or actual constant current source.
Would have very good drive.
You dont blast the first transistors at 30ma to get 17ma on the output drivers.
Would look at post #32 not more than 3ma is actually needed. most will use up to 8ma or 10ma.
The outputs would then bias at 1 to 3ma and you add the extra resistors as shown in #32 or many other better methods.
To bias the outputs at the needed 17 to 20ma for a 600 to 50 ohm load full swing. Not 8 ohms.
You need way way more gain with a Darlington or CFP output stage with a power transistor to get full swing into 8 ohms.
Or bias the living crap out of the outputs close to class A like many do, without using a Darlington or CFP to get full swing to a load
with poor gain.
By bootstrapping you mean tying the collectors to the emitters of the output transistors? Like so?
...actually, it improves the THD a little bit. I think it's because it helps remove base charges more quickly.
My designs are junk, so junk that i may just build an little 2 watt single ended amplifier but do it properly and once and for all.
I ran a simulation spanning several seconds. The THD is actually improving over time. It's pretty bad for the first 50ms after power up, but gets better as C1 gets fully charged.
I played around with it a bit. Interestingly enough, it did NOT perform better with constant current sources, so in the spirit of low component count, I put back the resistors. 30mA through the first stage was a bit excessive, so I lowered it to 15mA. This allows me to use the cheaper and smaller BD139/140. With MJE3055/2955 at the output I get about 0.01% THD @ 2.5W/1kHz. With this setup each transistor dissipates about 3.5W which may be a little excessive for some people's taste.
Does it make any diffrence to use the polcap model instead of the non polar cap model in ltspice?
OK. Final(?) version. 10mA through Q1 and Q2 and an added bias control. Roughly 0.003% THD. 3dB bandwidth about 600kHz. It might benefit from an extra layer of drivers in front of Q3 and Q4, but this is supposed to be minimalistic. I'm not in love with the third harmonic, but at -90dB from the fundamental, I'm not sure I should worry too much. I might draw up a PCB just for fun and I'm sure some part of my house "needs" a low to medium powered no-frills amplifier. 😀
Oh, and the bias current is 170mA. Seems like it needs to dip its toes pretty deep into Class A to perform well. That translates to about 5.5W power consumption at idle.
Another nice feature is that it has built-in overcurrent protection. At about 2A output current, R3 and R4 starve the output devices of base current.
A final remark: I had to put back the capacitor between the bases of the output devices to get an acceptable swing. I'm not sure I follow @WhiteDragon 's reasoning on why it would deteriorate after operating for a few seconds. It's not what I see in simulations or the breadboard prototype I built.
Those are terrible output transistors. Try current types such as MJW21193/MJW21194 or MJ21193/MJ21194. I've built a few that perform extremely well using modern parts.
It also helps to have 2EF designs, you need the extra current gain.
Use junk, get junk.
It also helps to have 2EF designs, you need the extra current gain.
Use junk, get junk.
Hi njswede,
I get it, but the quality of models for spice are critical and generally reflect above optimal performance. You need to include performance spread as well.
Sometimes (most times) if you know a circuit should work, you are much further ahead just building the darned thing. You can modify the prototype to fine tune it, and you get real, actual performance. You also can change out various devices to check how critical individual performance variations are, or matching for example. Plus you get a real feel for how it runs. Then if it works, you can listen to it or actually measure performance. Now there is truth.
I get it, but the quality of models for spice are critical and generally reflect above optimal performance. You need to include performance spread as well.
Sometimes (most times) if you know a circuit should work, you are much further ahead just building the darned thing. You can modify the prototype to fine tune it, and you get real, actual performance. You also can change out various devices to check how critical individual performance variations are, or matching for example. Plus you get a real feel for how it runs. Then if it works, you can listen to it or actually measure performance. Now there is truth.
I did build it on strip board yesterday and it performs well. Unfortunately, the only suitable transistors I had at home were TIP35/36 which aren’t great audio transistors either. But it showed decent THD and sounded OK to me.
I use SPICE mostly to catch mistakes and verify that a design isn’t broken. I’m fully aware that the results don’t always carry over to a physical build.
I use SPICE mostly to catch mistakes and verify that a design isn’t broken. I’m fully aware that the results don’t always carry over to a physical build.