Wanting a little gain

[cetoole]

Are you sure you got the output impedance correct? 3ohm just feels way too high, I would have expected close to that for the output stage shown before the application of feedback. Everything else feels reasonable.

 

[ruZZ.il]

wow she is gorgeous

 

[NelsonVandal]

Quote:

Originally Posted by cetoole /img/forum/go_quote.gif Are you sure you got the output impedance correct? 3ohm just feels way too high, I would have expected close to that for the output stage shown before the application of feedback. Everything else feels reasonable.

Yes I think so, both IRL and in simulator. As an example in LTSpice:
output voltage with no load = 1.1348 V, and with 10R load = 0.891091 V.
Zo=Rload(Vunloaded-Vloaded)/Vloaded = 10(1.1348-0.891091)/0.891091 = 2.73. I think the distortion figures indicates that this is right.

JISBOS unity gain simulates an output Z of 0.067R.

 

[FallenAngel]

NelsonVandal : That schematic sure looks like a JISBOS.

I do see the 33pF cap to ground is removed and 22pF added across feedback resistor, are you sure the 33pF is to be removed?

Looks like I should be able to do this mod quite soon, but I might not have those exact resistors around, I'll check tonight (when I'll be bringing the unit home from work).

Looking at the boards as pictured here, it looks as simple as replacing C3 (33pF cap with the 560R resistor), cutting the trace after it leading to output and replacing it with 2K2 resistor in parallel with 22pF cap. Piece of cake!

nikongod : The voltage between + and - per channel is around 60mV.

Thanks

 

[NelsonVandal]

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif NelsonVandal : That schematic sure looks like a JISBOS. I do see the 33pF cap to ground is removed and 22pF added across feedback resistor, are you sure the 33pF is to be removed?

There are more than one solution to get the reslut you want. As I understood it, you wanted as few changes as possible. With the original 1k input resistor, the capacitance of SK170/SJ74 should be enough to filter out RF. The value of the lag comp capacitor has to be choosen with regards to the resistor values, and 22 pF seems to do the job.

The amp could be slightly better balanced, faster, and the overshooting could be grinded down a bit by using lower resistor values and dual lag comp caps. By tampering with R1/R2 and R9/R10 (in my schematic) you could get a better voltage swing, but I thought I'd show a version with as few mods as possible.

 

[NelsonVandal]

Here's a more "optimized" version with better voltage swing, much lower noise, lower distortion, higher slew rate, less overshooting, higher open loop gain and much lower output impedance - 0.25 Ohm.

Pulse response, faster, less overshooting

Lower noise

Frequency and phase response

Open loop gain and phase (without the comp caps)

 

[NelsonVandal]

Distortion figures at 1 kHz, 1 V output

300R load

30R load

 

[FallenAngel]

Thanks Nelson, judging by how much better the optimized version measures, I guess I'll have to start fresh; keep the current setup as is (well, actually, sell it as is - it drives everything quite nicely) and build a balanced JISBOS with gain using optimized values. Looks like I don't have the extra tiny resistors for it and I don't want to tombstone everything with Dale resistors so I'll be ordering parts in a few days.

Thanks for the info.

 

[burgunder]

How much gain will the new circuit have?

 

[NelsonVandal]

Quote:

Originally Posted by burgunder /img/forum/go_quote.gif How much gain will the new circuit have?

The gain is 470/100 + 1 = 5.7

To FallenAngel, these aren't measurements, they're all simulations. I have measured my very messy amp using RMAA/soundcard, but I don't own an oscilloscope.

Regarding the "new" circuit, you have to measure the current through R9/R10, because it might behave different in real life depending on what transistors you use. I'm actually using those values 270R/10R and the current is about 7 mA through the BC550/560 and 3 mA through the JFETs. From a DC perspective LTSpice usually is correct. To me voltage swing was important since my amp's battery powered.

I must admit I haven't tried those feedback resistors and dual lag comp caps IRL, so it's a bit of a gamble

.