Wow... Just priced up parts for the tda4935 and they roll in at a huge £5 including postage.
Now if the bank of wife allows it, I will see if it will all fit in a tin somehow... Can i use normal RCA input and output, kinda like a super micro, micro amp. I have a toriod that gives 18v after rectifier so all i need is a tin and a fiver..... Sorry it may not be as technical as you guys are used to, but i dont have a clue how it all works, i just have a go and see
If i never post again... it went BAD!... real bad lol.
Now if the bank of wife allows it, I will see if it will all fit in a tin somehow... Can i use normal RCA input and output, kinda like a super micro, micro amp. I have a toriod that gives 18v after rectifier so all i need is a tin and a fiver..... Sorry it may not be as technical as you guys are used to, but i dont have a clue how it all works, i just have a go and see
If i never post again... it went BAD!... real bad lol.
Those are better, but AFAIK the MJLx302/x281 are the best in this category. The important parameter is not Cob but Cib. In the SPICE models it's Cje. Also Cbe. You want Cbe/Cje/Cib to be as low as possible for an output transistor. The actual size of Cbe is in the nF's. It is interesting to compare Cje in SPICE models you know are accurate.
Also, it is not the Ft at excursions that matters for crossover behavior but the Ft at Iq. I try to go for outputs with an Ft of over 20MHz at 100mA.
Cbe can also matter for small-signal transistors. Too much Cbe in the input stage will create a "glass ceiling" in amplifier BW that is not obvious. The BC5xx, MPSA18, 2N5210/5089 transistors have low Cbe. High-voltage transistors (japanese or not) tend to have larger Cbe and can actually lower BW if put in place of the others. So I shall proverbialize: Don't use high-voltage transistors for low-voltage applications.
Quasi-saturation and Vcesat are other parameters that get worse with higher Vcemax.
Also, it is not the Ft at excursions that matters for crossover behavior but the Ft at Iq. I try to go for outputs with an Ft of over 20MHz at 100mA.
Cbe can also matter for small-signal transistors. Too much Cbe in the input stage will create a "glass ceiling" in amplifier BW that is not obvious. The BC5xx, MPSA18, 2N5210/5089 transistors have low Cbe. High-voltage transistors (japanese or not) tend to have larger Cbe and can actually lower BW if put in place of the others. So I shall proverbialize: Don't use high-voltage transistors for low-voltage applications.
Quasi-saturation and Vcesat are other parameters that get worse with higher Vcemax.
The MJL/NJL might be overlarge. I think that the MJW/NJW are slightly smaller. Onsemi didn't seem to have much in the way of extra nice to220 outputs, however. . .
Can sturdy drivers be output devices for a little amp?
MJE15028/MJE15029
MJE15030/MJE15031
MJF15030/MJF15031
Edit:
I see that Kean mentioned them 4 minutes ago.
Edit2:
However, I wonder if the much lower voltage Tip31a Tip32a would be lower loss, more placid (easier for stable), and cooler running?
Can sturdy drivers be output devices for a little amp?
MJE15028/MJE15029
MJE15030/MJE15031
MJF15030/MJF15031
Edit:
I see that Kean mentioned them 4 minutes ago.
Edit2:
However, I wonder if the much lower voltage Tip31a Tip32a would be lower loss, more placid (easier for stable), and cooler running?
Last edited:
I'm also doing a chip project, just as basic as possible and barely meeting the 9w+9w parameter. Since mine is for use with the computer, I'm thinking of installing 1/8" stereo jack so it is really easy to fit in the tin.
Last edited:
I don't know why that would be the case. The TIP transistors are very old, they must be a different process. I would save them for lightbulb flashers when better types are easily available.
Probably, the MJE15028/MJE15029 (or little FJPF5200/FJPF1943) could give some very good (lower noise) performance in the no bias setup. I think they're worth a try. One of the projects here in the chip amp forum uses 5200, 1943 pushed by a TDA729x chip and I think those outputs are used without bias. It seems that the negative feedback was enough for noise reduction from such fine outputs? I don't quite know the particulars.
Last edited:
I've had another play with the circuit in real life - I tried both EF outputs and CFP. I think I prefer the performance of the CFP. One thing is the superior bias stability, which is likely needed when your heatsink is a tin of golden virginia
I'm getting some parasitic oscillations but the circuit is a mess of wires and jumpers on a breadboard. Next week I will have time to etch a PCB and see how the circuit really performs. I will test it with the TIPs and see what the actual measured performance is like. I will post harmonic spectra.
This is not to say I haven't taken aboard the recommendations for better output transistors, I just want to see how well the TIPs will actually work in real life. A 100W amplifier I repaired in real life seems to achieve rather good performance with "glacial" 1Mhz Ft output transistors.
I'm getting some parasitic oscillations but the circuit is a mess of wires and jumpers on a breadboard. Next week I will have time to etch a PCB and see how the circuit really performs. I will test it with the TIPs and see what the actual measured performance is like. I will post harmonic spectra.
This is not to say I haven't taken aboard the recommendations for better output transistors, I just want to see how well the TIPs will actually work in real life. A 100W amplifier I repaired in real life seems to achieve rather good performance with "glacial" 1Mhz Ft output transistors.
Last edited:
You are going to confuse people by referring to glitching as noise.
When you go totally without bias, you not only have transistor glitches to worry about, but the current spike induced by the error voltage. On miller compensated amps this will be monstrous. It will be near impossible to achieve low distortion this way. AC behavior will change dramatically with output.
When you go totally without bias, you not only have transistor glitches to worry about, but the current spike induced by the error voltage. On miller compensated amps this will be monstrous. It will be near impossible to achieve low distortion this way. AC behavior will change dramatically with output.
I'm not going to be running the amp with no bias, I just turned it right down to see how the circuit coped with the glitch - apparently quite well.
We mustn't get carried away
This is an amplifier in a tin. I think the fact that I'm putting a mini-blameless in a tin is pushing it already How out of date would it be to use a an IR4312 to feed power to a complimentary class A output stage? I wonder if the single channel version IR4311 could be used with asymmetrical power supplies, to save switching loss. That would require more chips though, which I guess are already breaking the rules.
Last edited:
Assuming you could find some output transistors that were still linear with 5 volts across them, the minimum 1A bias current would still be burning over 20 watts for a pair of amplifiers. Getting rid of that much power in that small of a package would require a high speed fan that would make enough noise that you'd be a lot better off with the D amp chips by themselves.
Chips are okay to use, and observe that this is posted in a chip amp forum.
I'm aware that you can't safely run 9w bias (18 watts bias for stereo) in a mint tin . There's hugely generous leeway for Classes Aa, AaH, AaG and other dynamic Class A. Dynamic class A that also uses adaptive power source is an extremely efficient state of the art solution that may be a future for linear audio.
There's just 2 things to watch out for:
Please don't convert the analog input signal to digital inside of the mint tin (providing an "extra" digital input is fine if the amp also has a source selector to allow the possibility of an all analog path).
The Dynamic class A amplifier (Aa, AaH, AaG, etc. . .) must pass common tests at least as nicely as Class B for the first 9 watts per channel, stereo.
For AaH, that looks like, Class A tiny bias, class a 9w (adaptive bias), but class H for headroom management at points over 9w. Right?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.