Hi charles,
I think I speak on behalf of, myself, when I say, you ruined the fun! Thanks all the same, big help.
How about making the pcb adaptable to full bridge? Provisions for 2 more output transistors off the comparator, seperate driver as discussed before, not sure how that affects the feedback take off though. Just a thought.
I think I speak on behalf of, myself, when I say, you ruined the fun! Thanks all the same, big help.
How about making the pcb adaptable to full bridge? Provisions for 2 more output transistors off the comparator, seperate driver as discussed before, not sure how that affects the feedback take off though. Just a thought.
Ouch !!!😱
If this is done in a clever way then it would be possible to just leave away the 2nd half of the bridge when making the PCB. I also once had ideas about a 2nd module that is piggybacked on the main one and which is only carrying the second half of the bridge (i.e. no modulator nor any other input circuitry).
Regards
Charles
Yeah just what I had in mind, as long as it wouldn't affect the feedback loop or anything? Would be neat if it could be built up like that without much effort or re-design anyway, something to kick around huh? It would add nothing to the original to make provisions for it anyway, at least nothing worth mentioning.
If you build the control loop such that it senses differentially (either between output and gnd or out+ and out-), the same loop could be wired to drive either fb or hb. The control circuit on my half and full bridge amps is exactly the same for both (different from the example ckt in the patent).
It is best to make sure the whole loop is differential (so not a diff amp followed by an unbalanced ckt please. Ever since Harris did that in their 4080 demo board it's been popping up in experimental designs time and again).
Differential input comes for free as well.
It is best to make sure the whole loop is differential (so not a diff amp followed by an unbalanced ckt please. Ever since Harris did that in their 4080 demo board it's been popping up in experimental designs time and again).
Differential input comes for free as well.
O.K. I admit everything already comes for free with your topology.
But what is to say against the Harris topology of feedback takeoff, apart from having active components in the feedback path (which I don't like either since they contribute to nonlinearity rather than being cancelled by the feedback loop) ?
Or have I given your answer by myself already ?
Regards
Charles
Wellll- in the harris case the feedback was taken entirely before the output filter. Except with very unusual circuitry one hasn't got a snowball's chance in hell to get good linearity and cmrr with this sort of signal!
The upshot is indeed that the diff amp is going to add substantially to distortion. After that, ground-referred circuitry close to a power stage isn't going to help either.
The upshot is indeed that the diff amp is going to add substantially to distortion. After that, ground-referred circuitry close to a power stage isn't going to help either.
I was thinking of using a burr brown instrumentation amp for the differential input....should get a fairly clean signal out of that? Gain is adjustable with one external resistor. They've got a few of em.
Oh, apparently I haven't managed to argue against diff amps clearly enough 
An inst amp certainly won't fly. It'll need to handle the entire output swing of the output stage. Both inputs and outputs are limited to the power supply of the inst amp. Furthermore you have to implement loss, not gain. Inst amps owe their spectacular CMRRs to the fact that the gain is made differentially, with only the actual difference amp requiring precision resistors.
The HF requirements are another can of worms which I'll leave closed.
An inst amp certainly won't fly. It'll need to handle the entire output swing of the output stage. Both inputs and outputs are limited to the power supply of the inst amp. Furthermore you have to implement loss, not gain. Inst amps owe their spectacular CMRRs to the fact that the gain is made differentially, with only the actual difference amp requiring precision resistors.
The HF requirements are another can of worms which I'll leave closed.
Bruno,
i'm still not sure, what is more preferable FQP13N10 or FDP3682 (let's even assume that body diodes are equal)?
WOW!!!
...Hi ...!
Thanks for pushing me to this this thread.
Now I have three options:
- Find a nice PWM-controller ( ...growing doubts about that...)
- Designing my own Carrier PWM Control .. 🙄
- Investigating a UCD-clone
I will definitely have a look to a simplified UCD simulation.
..read something about a Muena Chip with adaptive dead time control.
....is this similar to the adaptive dead time control of the LM5104, which I promoted in the Class-D-Distorsion-Thread?
Hm, I am not sponsored by NATIONAL, also not an employe of them
(and so I think putting the link two times is sufficient ),
but I still think that the LM5104 looks promising.
It has an detection for the Vgs and drives the MosFet through
a traditional diode + resistor parallel connection.
For turn off the max. sinking current of the chip is pulling down the
MosFet quite fast and the chip internal control circuit can detect the
Vgs of the MosFet. Speed of MosFet turn on can be adjusted by the
resistor. If we select the proper value then we can provide a real good chance for the body diodes of the MosFets to survive. 😉
Are any theoretical comments or even better experiences about that chip available here?
There was a discussion about a coreless choke for the outputfilter.
I think the only suitable geometry for this would be a torroid.
You could use a pertinax torroid (or any flame retarding material, which is non conductive and non magnetic).
The torroid geometry forces the entire field inside the torroidal shape, even if there is no material with high permeabilty.
Well, in reality we will always have some leaking field, but with that shape it should be comparably low, if the windings are distributed uniformly around the entire torroid.
Nevertheless, I do not think that a coreless choke is the best solution (...just a guess, not calculated/tested).
Due to the low permeabilty of the air, we would need much more turns than with a gapped core design. I would not expect that
we can beat the losses ( ==> Q) and size of an optimized gapped core design with a coreless design.... but might be worth a trial, may be it is good enough and would save the core.
The current and the magnetic field will have a high frequency component and a signal component. High frequency component will
be there continously but at lower level than signal currents. The signal currents will have much higher max. values, but will only occure according the music programm. Choice of wire is a trade off.
Probably we would not need to use real thin stranded HF-litz.
Hopefully multistranded wires in the range between 0.2mm to 0.3mm per strand might work.
Bye
Markus
...Hi ...!
Thanks for pushing me to this this thread.
Now I have three options:
- Find a nice PWM-controller ( ...growing doubts about that...)
- Designing my own Carrier PWM Control .. 🙄
- Investigating a UCD-clone
I will definitely have a look to a simplified UCD simulation.
..read something about a Muena Chip with adaptive dead time control.
....is this similar to the adaptive dead time control of the LM5104, which I promoted in the Class-D-Distorsion-Thread?
Hm, I am not sponsored by NATIONAL, also not an employe of them
(and so I think putting the link two times is sufficient
),but I still think that the LM5104 looks promising.
It has an detection for the Vgs and drives the MosFet through
a traditional diode + resistor parallel connection.
For turn off the max. sinking current of the chip is pulling down the
MosFet quite fast and the chip internal control circuit can detect the
Vgs of the MosFet. Speed of MosFet turn on can be adjusted by the
resistor. If we select the proper value then we can provide a real good chance for the body diodes of the MosFets to survive. 😉
Are any theoretical comments or even better experiences about that chip available here?
There was a discussion about a coreless choke for the outputfilter.
I think the only suitable geometry for this would be a torroid.
You could use a pertinax torroid (or any flame retarding material, which is non conductive and non magnetic).
The torroid geometry forces the entire field inside the torroidal shape, even if there is no material with high permeabilty.
Well, in reality we will always have some leaking field, but with that shape it should be comparably low, if the windings are distributed uniformly around the entire torroid.
Nevertheless, I do not think that a coreless choke is the best solution (...just a guess, not calculated/tested).
Due to the low permeabilty of the air, we would need much more turns than with a gapped core design. I would not expect that
we can beat the losses ( ==> Q) and size of an optimized gapped core design with a coreless design.... but might be worth a trial, may be it is good enough and would save the core.
The current and the magnetic field will have a high frequency component and a signal component. High frequency component will
be there continously but at lower level than signal currents. The signal currents will have much higher max. values, but will only occure according the music programm. Choice of wire is a trade off.
Probably we would not need to use real thin stranded HF-litz.
Hopefully multistranded wires in the range between 0.2mm to 0.3mm per strand might work.
Bye
Markus
Bruno Putzeys said:Oh, apparently I haven't managed to argue against diff amps clearly enough
An inst amp certainly won't fly. It'll need to handle the entire output swing of the output stage. Both inputs and outputs are limited to the power supply of the inst amp. Furthermore you have to implement loss, not gain. Inst amps owe their spectacular CMRRs to the fact that the gain is made differentially, with only the actual difference amp requiring precision resistors.
The HF requirements are another can of worms which I'll leave closed.
Hmmm, so what you're saying is I shouldn't buy a module from Jan-Peter? From page 7 of his UCD pdf file:
"For this reason, the UCD cell is equipped with differential inputs. These are used either to receive the input
signal with reference to the local ground of the preamplifier circuit (right), or to build a fully balanced
input in “instrumentation amplifier” fashion (left)."
Sure looks like an instrumentation amplifier to me.
Hi Chris,
A small correction: The UcD pdf, what you can download from our website www.hypex.nl, is the original UcD information written by Bruno..........
The small ucD180 is available right out of stock at the moment.
Regards,
Jan-Peter
www.hypex.nl
A small correction: The UcD pdf, what you can download from our website www.hypex.nl, is the original UcD information written by Bruno..........
The small ucD180 is available right out of stock at the moment.
Regards,
Jan-Peter
www.hypex.nl
Hi Jan, So that's Bruno trashing his own work? LoL..I think we had a misunderstanding regarding the instrumentation amp..
I'm alot more interested in your 400W, any news, special request only or..it's on my xmas list!
I'm alot more interested in your 400W, any news, special request only or..it's on my xmas list!
Hi Chris,
The use of this type of comparator has a big advantage of also making very easy a symmetrical input. Why shoudn't we use this?
Bruno is the designer/invertor of the UcD concept, mine company may use UcD techonlogy for making Class-D amplifiers. So I also used the technical pdf information about the smallest UcD amplifier, however this documention is written by Bruno / Philips. It's a funny coincident Bruno is on this Forum, and can see his own documentation back....
The UcD400 is nearly ready for putting in production, in about 6-8 weeks this version will be deliverable.
Regards,
Jan-Peter
The use of this type of comparator has a big advantage of also making very easy a symmetrical input. Why shoudn't we use this?
Bruno is the designer/invertor of the UcD concept, mine company may use UcD techonlogy for making Class-D amplifiers. So I also used the technical pdf information about the smallest UcD amplifier, however this documention is written by Bruno / Philips. It's a funny coincident Bruno is on this Forum, and can see his own documentation back....
The UcD400 is nearly ready for putting in production, in about 6-8 weeks this version will be deliverable.
Regards,
Jan-Peter
Hi,
Didn't make sense to me either it seems the perfect circuit for an input stage, I think Bruno thought I was thinking about using it inside the feedback loop? I'm not sure, I think there was a small misunderstanding, I'm sure he'll clear it up when he's around.
The subject prior to me mentioning using a burr brown instrumentation amp for the input was differential feedback loops so I can see how it happened if that's indeed what did occur.
Thanks for the heads up on the 400, I'm looking forward to it. Might make up the power supply for one soon in fact.
Regards,
Chris
Didn't make sense to me either it seems the perfect circuit for an input stage, I think Bruno thought I was thinking about using it inside the feedback loop? I'm not sure, I think there was a small misunderstanding, I'm sure he'll clear it up when he's around.
The subject prior to me mentioning using a burr brown instrumentation amp for the input was differential feedback loops so I can see how it happened if that's indeed what did occur.
Thanks for the heads up on the 400, I'm looking forward to it. Might make up the power supply for one soon in fact.
Regards,
Chris
classd4sure said:
Oh you mean on the input side? That's fine of course. I was reacting to someone who wanted to take feedback from a full bridge amp using an instrumentation amp from burr-brown. Read that post concerning signal levels and gain again.
Building an inst amp using 2 op amps and an UcD with a symmetrical modulator is quite a different thing. Since I "invented" that scheme (=a UcD as the diff amp in an inst amp configuration) I wouldn't argue against that. I couldn't if I wanted to, because I can't see anything wrong with it.
More careful reading prevents silly posts.
Bruno Putzeys said:
Of course, if I wrongfully construed your question regarding the BB amp as wanting to put it in the feedback loop, the same comment goes for me too.
Hi,
You certainly did, but I'll forgive ya. I had said "for the differential input", but it was kind of a throw in the topic, these things happen on forums, no harm done at all. I've a pretty good idea what kind of op amp to use for anything going into the feedback loop.
Regarding the input stage now, I liked the looks of those burr brown instrumentation amps, only takes one, the high precision resistors are all internal, only takes one external resistor to set the gain on it. It's kind of hard to make a "good" instrumentation amplifier I thought that would make life easy.
Good news is I got my power supply going, 16 volt rails under load (10 ohm), 19V rails with no load. Should be perfect for a protoboard test circuit.
Now for the question..
I popped off a small toroidal transformer from a PC PSU, thinking I could use it for the output filter. I can't cut a gap in it, any thoughts? Not really sure how to go about that, I haven't measured it yet either. At this stage of the game I'm not aiming for perfect, just something workable. It's that or an air core I'm afraid.
Regards
Chris
You certainly did, but I'll forgive ya. I had said "for the differential input", but it was kind of a throw in the topic, these things happen on forums, no harm done at all. I've a pretty good idea what kind of op amp to use for anything going into the feedback loop.
Regarding the input stage now, I liked the looks of those burr brown instrumentation amps, only takes one, the high precision resistors are all internal, only takes one external resistor to set the gain on it. It's kind of hard to make a "good" instrumentation amplifier I thought that would make life easy.
Good news is I got my power supply going, 16 volt rails under load (10 ohm), 19V rails with no load. Should be perfect for a protoboard test circuit.
Now for the question..
I popped off a small toroidal transformer from a PC PSU, thinking I could use it for the output filter. I can't cut a gap in it, any thoughts? Not really sure how to go about that, I haven't measured it yet either. At this stage of the game I'm not aiming for perfect, just something workable. It's that or an air core I'm afraid.
Regards
Chris
Gapped toroid.
IMO using an IC inst amp to convert the signal to single-ended and to proceed in a ground referenced manner is always going to produce more problems than holding on to differential signalling all the way, even if the total CMRR is less than that of the inst amp. The point is that after the inst amp, CMRR is zip.
Gapping toroid cores: Dremel sells 1mm thick grinding discs for their hand-held drills. Mount one of these on a drill stand and run it at low speeds (250RPM or so). You can grind through 4mm of ferrite in about 5 minutes. It takes patience and a steady hand but the results are good.
IMO using an IC inst amp to convert the signal to single-ended and to proceed in a ground referenced manner is always going to produce more problems than holding on to differential signalling all the way, even if the total CMRR is less than that of the inst amp. The point is that after the inst amp, CMRR is zip.
Gapping toroid cores: Dremel sells 1mm thick grinding discs for their hand-held drills. Mount one of these on a drill stand and run it at low speeds (250RPM or so). You can grind through 4mm of ferrite in about 5 minutes. It takes patience and a steady hand but the results are good.
- Status
- Not open for further replies.