An obvious way to simplify the design is if you were to remove the bootstraps in the output stage. That will save 7 components. But then that would cost some efficiency, as the dropout voltage will rise 4 V. You can also remove the first amplifier stage, saving 4 components, but that will cause some DC problems, related to impedance of the driving stage.
I think the modulation of T15 is only of very low significance. Neither T9 or T5 are sensitive to change in their collector current through R24, the delta voltage changes only by a value determined by the Re (around 25 Ohms). So any modulation from E of T9 (in the region 0,5V) will be dampened 4000 times before it is sent back in the input stage in common mode. Here it results in a microscopic amount of Negative Feedback. I think no problem.
I think the modulation of T15 is only of very low significance. Neither T9 or T5 are sensitive to change in their collector current through R24, the delta voltage changes only by a value determined by the Re (around 25 Ohms). So any modulation from E of T9 (in the region 0,5V) will be dampened 4000 times before it is sent back in the input stage in common mode. Here it results in a microscopic amount of Negative Feedback. I think no problem.
Hi lars. Nice .. really nice ...
Still if i would you i would leave the input cap in the circuit. without it and a low inpedance load you will get a serious output DC offset shift, when connecting the source, with such a high resistance seen by the -IN on the diff. input. i get like 400 - 500mV maybe more with my calculation when running from a +/35V supply.
With the input cap, i see no problem. 🙂
Still if i would you i would leave the input cap in the circuit. without it and a low inpedance load you will get a serious output DC offset shift, when connecting the source, with such a high resistance seen by the -IN on the diff. input. i get like 400 - 500mV maybe more with my calculation when running from a +/35V supply.
With the input cap, i see no problem. 🙂
Lars,
Looking good, and very generous of you!
Of course there are things I'd do differently - but hell, we're all different - and there are several ideas in this which I believe are very good.
Have you built it, and how does it sound?
I guess that's what matters.........😀
Cheers,
Hugh
Looking good, and very generous of you!
Of course there are things I'd do differently - but hell, we're all different - and there are several ideas in this which I believe are very good.
Have you built it, and how does it sound?
I guess that's what matters.........😀
Cheers,
Hugh
Not all of it yet .. but the essential parts i have built (as a birds nest), when i can get a set of PCB's produced, i will post what the sound is like.
If anybody else beats me to it, please also post.
AKSA: Any new suggestions are of course welcome. I am sure you can make some good contributions. And since this will be an open source project, everybody in here can benefit.
Anybody would like the original schematic in Corel Draw, i can post it in a ZIP file. It's easier to modify than start over 🙂
If anybody else beats me to it, please also post.
AKSA: Any new suggestions are of course welcome. I am sure you can make some good contributions. And since this will be an open source project, everybody in here can benefit.
Anybody would like the original schematic in Corel Draw, i can post it in a ZIP file. It's easier to modify than start over 🙂
AKSA said:
Yes there are some refreshing idears in it. And i like the ULCA2 more than the ULCA1, it will have a more even slewrate in each directions.
At +/- 35V you get:
With the bootstraps: 60.4W in 8 Ohms.
Without the bootstraps: 45.8W in 8 Ohms.
The cooling demand is the same for both examples.
Soundwise the difference may not be significant.
Have a good weekend all!
With the bootstraps: 60.4W in 8 Ohms.
Without the bootstraps: 45.8W in 8 Ohms.
The cooling demand is the same for both examples.
Soundwise the difference may not be significant.
Have a good weekend all!
i would think a low cost amp might want unusually good psrr since regulated supplies would be out of the question
in this example you have inferior psrr, particularly with the resistive bias of R18,28 for T12,13;
a quick est of T12,13 psrr at the output can be had from 26 mV/ 35 Vsupply *1% tol(R28,18) * 50 Av ~= 1/2800 or only 70 dB
while you might argue R28&18 will likely match better than 1% I would suggest you consider the delta R spec for exposure to soldering temp and the variation due to hand assembly (I regularly and severely overcook parts when reworking smt boards)
in this example you have inferior psrr, particularly with the resistive bias of R18,28 for T12,13;
a quick est of T12,13 psrr at the output can be had from 26 mV/ 35 Vsupply *1% tol(R28,18) * 50 Av ~= 1/2800 or only 70 dB
while you might argue R28&18 will likely match better than 1% I would suggest you consider the delta R spec for exposure to soldering temp and the variation due to hand assembly (I regularly and severely overcook parts when reworking smt boards)
Hi jcx,
I only understand your remark half-way.. did you considder that the humm introduce into T12/T13 is a common-mode humm? Will it be rejected by the CMRR or the amp?
Regards,
Thijs
I only understand your remark half-way.. did you considder that the humm introduce into T12/T13 is a common-mode humm? Will it be rejected by the CMRR or the amp?
Regards,
Thijs
that's where R tol enters the equation, if R28,18 aren't exactly equal there is dfferential signal at the diff pair input
I'm not sure what advantage this arrangement has for the differential input pair rather than an all PNP Darlington configuration which would eliminate R18 and R28.
or a CFP for Diff-input-pair .. I have seen that before...
ellimination R18/R28 would bias T12/T13 for too low Iq.... Need atleast a couple hunderd uA to get Igain up?
ellimination R18/R28 would bias T12/T13 for too low Iq.... Need atleast a couple hunderd uA to get Igain up?
I don't mean just get rid of R28/R18 but replace T12 and 13 with PNP devices then R28/R18 can disappear🙂
Hi Richard C,
if we would replace T12 and 13 with PNP devices (emmirot connected to bases of 'real' diff-pair.. same thing would happen.. only the base-current of the 'real' diff-pair is the emmitor current of T12/T13 .. that could be a to low a current... 😕
Regards,
Thijs
if we would replace T12 and 13 with PNP devices (emmirot connected to bases of 'real' diff-pair.. same thing would happen.. only the base-current of the 'real' diff-pair is the emmitor current of T12/T13 .. that could be a to low a current... 😕
Regards,
Thijs
True, Ic for T12/13 if PNP would only be a few uA but this can easlily be increased with resistors (10k perhaps) across T10/11 base-emitters and this current would be constant regardless of supply voltage.
Hi
Yes this would definitely work, and the collectors should go to V- directly (not to the C of T10 and T11) so the base current would not go into the signal chain.
It is a bit hard to see the cost / benefit comparation of these two solutions. Both would work fine.
A completely symmetrical voltage amp is also a possibility, but it will cost another 4 transistors.
Yes this would definitely work, and the collectors should go to V- directly (not to the C of T10 and T11) so the base current would not go into the signal chain.
It is a bit hard to see the cost / benefit comparation of these two solutions. Both would work fine.
A completely symmetrical voltage amp is also a possibility, but it will cost another 4 transistors.
Hi Lars, what about the suggested improvement in PSRR resulting from eliminating R28/18 do you think this is significant? Also, would there be any benefit in replacing T5/7 with a current-mirror? I realise that these are essentially different ways of doing the same thing so there might not be a definitive answer.
Interesting amp though and miniaturised too. Finding matched pairs of those tiny transistors could be fun🙂
Interesting amp though and miniaturised too. Finding matched pairs of those tiny transistors could be fun🙂
So i'm not so familiar with use of depletion mode FET's as current sources. It seems to me that there is typically a very large Vgs off tolerance on these devices, so how do you determine the value of the current source without adjustment?
And how about high temperature leakage of the FET, from datasheets it shows that the leakage goes from 10uA to 1 mA at 125 deg. C. Is that really the case?
If use of these 'Current Diodes' can improve the design, then why not?
And how about high temperature leakage of the FET, from datasheets it shows that the leakage goes from 10uA to 1 mA at 125 deg. C. Is that really the case?
If use of these 'Current Diodes' can improve the design, then why not?
- Status
- Not open for further replies.