UCD Based Fully Differential Full Bridge 450W RMS Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Konrad said:
This seems to work!

Basic circuit without classd4sure's input stage.

The input/controll-cuircuit shown by classd4sure would improove performance as open loop gain also in higher.

Abt the open loop gain have someone tried higher open loop gain?
I mean like replacing the resistors R121 and R122 with smal caps in Classd4sure's input stage ?

Yeah I had tried that, wasnt' able to get it to converge, doesn't mean it can't work of course. Also tried a cap parallel with the feedback resistors you mentioned .... and in series...sadly, non converged.
 
Hi,

Kind of hard to follow it but..

The diodes in the comparator will pass no current.... it will actually "compare" with them totally removed, but not as good. I'll go out on a limb by saying this but I think they add a small measure of hysteresis to it which it seems to like. I've changed them to schottky's as well for the "hot carrier" aspect, trr characteristic.. faster and cleaner switching I thought. Maybe someone more qualified can comment on that.

What's the purpose of the ground between RC going to Q13?

That should be a small signal schottky diode for the baker clamp.

15V for the drivers will add a bit more delay as JohnW was saying, with 12V drive you get about 10Volts at the gate which is perfect to fully enhance, I've tried 15V as well with every other version of the circuit and it was always far worse, I guess because of that added delay.

I'm messing around with the driver stages tonight, see how far I get with that, will also try to implement your balance circuit.
I think JohnW had an excellent suggestion with the added follower.

We'll compare notes ?

How is shoot through on it, I've found it difficult to balance things to minimise it, and I'm wondering how it is with your balance circuit added.

Glad to see you're still at it!

Regards,
Chris
 
What's the purpose of the ground between RC going to Q13?
its for the balance if impedances seen by the bases of the common mode transistors. ; the sorce resistanse seen by the bases of Q13 and Q14 should bee equal. The first path is the paralell R18, R19, the 2nd is R20 wich should bee 5k. ( i set it to 4.7k as it is standard that i have somwere in my draver :clown: ).

ABT driving voltages and diodes, yes/ well i havnt payd much attention to them yet. Thanks for comments.
 
classd4sure said:
Hi,

Kind of hard to follow it but..

The diodes in the comparator will pass no current.... it will actually "compare" with them totally removed, but not as good. I'll go out on a limb by saying this but I think they add a small measure of hysteresis to it which it seems to like. I've changed them to schottky's as well for the "hot carrier" aspect, trr characteristic.. faster and cleaner switching I thought. Maybe someone more qualified can comment on that.

What's the purpose of the ground between RC going to Q13?

That should be a small signal schottky diode for the baker clamp.

15V for the drivers will add a bit more delay as JohnW was saying, with 12V drive you get about 10Volts at the gate which is perfect to fully enhance, I've tried 15V as well with every other version of the circuit and it was always far worse, I guess because of that added delay.

I'm messing around with the driver stages tonight, see how far I get with that, will also try to implement your balance circuit.
I think JohnW had an excellent suggestion with the added follower.

We'll compare notes ?

How is shoot through on it, I've found it difficult to balance things to minimise it, and I'm wondering how it is with your balance circuit added.

Glad to see you're still at it!

Regards,
Chris


YPP ! There is definetly som more attention abt the clamping to bee done, as the shot thru in MKII is high(!) the implement of "Bruno clamping" and some value adjustments/rearrangement, it became much better.

Also adjusting the Ton time ( switch losses then neads attention) by adjusting R11, R12, R14 andR16.
i belive we will end up with somthing like Toff<Ton+(R*Cin) choose biger than minimum R will increase losses caused by exstended Ton. At the same time bigger R will reduse shot thru, here is definetly some energy / timing to take care of, after the clamping net is improoved.

But with better or 'Bruno clamping' and 'minor' adjustments in the resistor network arrangement is the place to improove the circuit "ballansed comparator with added feedback" as i see it as a unit behavior. In a different tread under Digi amp there were som suggestion on a simplified way to calculate the fiter/ and outer timing and internal delays.

Abt lower driving voltages for the output dev ; Yes lover voltage is a must! there is in some datasheats a internal zener gate-sorce it's important not to damage or overdrive, current limited by R, V drive and 'internal-zener'.
 
Uh! these output devices ! :bawling:

Well after a little fidling;
Different clamping- "Bruno clamping"
Timing controll to remoove shot thru; Ton>Toff
Feed forward switch diodes in pre switching stage.

It simulates ok, a little bit slow maybee. :smash:
 

Attachments

  • mkiii copy.gif
    mkiii copy.gif
    24.8 KB · Views: 1,336
Am i writing all for my self?

Redusing the ripple on the output, using different filtering, and NOT taking care of the phenomenas in the output filter itself.



I belive it is abt ready for a PID control loop to take care of the outer loops. Or maybee someone has a better idea?

At this stage things are getting werry interesting!
 

Attachments

  • mkiii- a.gif
    mkiii- a.gif
    27.6 KB · Views: 1,263
I miss the "lead" part in the NFB on the latest incarnation of your circuit.
BTW: Even if a simulation looks better when taking feedback from a different filter than the one the load is connected to, I wouldn't do it. NFB doesn't "see" the real output signal in this case but a "varnished" one.

Regards

Charles
 
Hi,

Is common mode feedback truly a necessity? It's my understanding a truly "differential" circuit does a fine job of rejecting common mode voltage, and it certainly seems to.

Granted if you can get a decent cmfb working well, great, but I don't feel it truly requires it. Rejection can be further improved by a simple change to the output filter as well.

Regards,
Chris
 
If you use a single differensial to handle a bridged output. AND you tend to use outer coltroll amps what will happen when thiese have great offsett on their output?

If your using two differensial stages and then bridge, its directly comparable to two separate amplifiers in bridge konfiguration. (like bridging UCD 180)

With one differensial pam you must have some way to avoid the controllers go into klipping. (nested feedback loops)

Else it should not matter much if there is some common mode on the output. Also this form of common mode rejection is NOT the werry best, and i belive just abt sufficient enough.
Thou it can bee improoved by moving the feedback to common base but with more components involved.

Dependent the rest of the circuit, as it is only a part of a hole system, and requierements...


HOW to improove common mode in the output stage filtering?

Ronny
 
classd4sure said:
Hi,

Is common mode feedback truly a necessity? It's my understanding a truly "differential" circuit does a fine job of rejecting common mode voltage, and it certainly seems to.

Granted if you can get a decent cmfb working well, great, but I don't feel it truly requires it. Rejection can be further improved by a simple change to the output filter as well.

Regards,
Chris

You are right, but its not the output filtering handling this. Audio band common mode, its the suppression of such happening in the audio band. That has to bee taken care of in a simpler way.

JohnW said:
Hi Chris & Konrad,

Nicely done to both of you - a good starting point - and clever CM balance Konrad :)

Output stage needs a bit more work - to make it faster & hence reduce the switching losses (how about adding a EF buffer with say FMMT617 / 717) - you could also add feedforward from the buffered driver to reduce the effects of circuit capacitances in the level shifter section.

John


And this common mode insert was not so klever after all, there is a much simpler way to cope with it [for a fraction af a second i thougt i was a genius :bawling: ]. And it definetly simplifyes things :D .

All there is to do is turning the current generator circuit in the classD4shore's org post inside out, and replace the transistor with a npn type.

Then the input stage is with all npn eksept the first differensial stage. How abt mos driver in NPN only konfig just for funn?
 
Konrad said:


You are right, but its not the output filtering handling this. Audio band common mode, its the suppression of such happening in the audio band. That has to bee taken care of in a simpler way.




And this common mode insert was not so klever after all, there is a much simpler way to cope with it [for a fraction af a second i thougt i was a genius :bawling: ]. And it definetly simplifyes things :D .

All there is to do is turning the current generator circuit in the classD4shore's org post inside out, and replace the transistor with a npn type.

Then the input stage is with all npn eksept the first differensial stage. How abt mos driver in NPN only konfig just for funn?

Hi Konrad,

Thanks for the further explanation..

I came to the same conclusion about your first attempt at a balance circuit, and have been thinking of making a whole new comparator.. :whazzat:

So this idea of yours interests me, do you care to post the schematic?

I don't think you can do an all npn driver stage.. have to use common collector followers for their high current drive ability and low output impedance. So far I haven't attained better performance for the driver than the very simplistic one posted originally.. but I'll see how it works with capacitive coupling next.

Regards,
Chris
 
Hi Ronny,

I think by doing that to the current source you starve the comparator ... effectively I believe it creates a second oscillator between the top half of the comparator and the bottom half, each working against each other. You can achieve the same effect just by replacing the 1k resistor in the current source with another diode in ... anti series, as a diode OR gate connected to the base.

While it does completely eliminate common mode voltage.. it no longer modulates the signal very well at all... and so the work continues.

Regards,
Chris
 
Thanks reply.
There is two Things neaded a closer look;
The Ton/Toff at the mosfets, if i want to have plenty safetymargin and not going for a shot thru. If i go for a single resistor on SWITCH ON pass there will bee to much switch losses and less losses givs more shot thru.
This can bee solved by placing a smal delay before the first "SWITCH". Delay by R30 R31 C11, C12. hence redusing switch losses at the next stage is a possibility.
Someone mentioned somthing abt dv/dt limit at the gates of the mos, this i have litle knowlege abt. But this may sertainly bee somthing to bee avare of.

Switching the mosfets of hm there is a smal coil in the lead and... better protect the driver by reverse diode D218, D219, D220, D221.

Reconfiguring again! and this time to bistabil toggle, thus having two toggles in series.
Remember Bruno saying there should bee another 3 components in the circuit loading the input current, and all is passive R24, R25, R26 makes the transistors here Q5 and Q6 act like a bistabil toggler.

The C2, C3, C4, C5 and R3, R4, R6, R7 is my estimate for the noise killer around the output switches.
Its abt time to start prototyping:devilr:
Maybee i should make som pcb's as well :D

someone claiming PCM triangle modulation microcontroller stuff etc.... well i prefer this simply of beauty!

Thou its still simplifyed, and the rest of it ?

I hope someone likes it !
 

Attachments

  • mk-iv.zip
    28.5 KB · Views: 409
Hi,

I've re-optimised my circuit, and made a few very minor changes.

I should add that I've also tried numerous variations with the driver stage, it remains the same because nothing else has matched its performance, and so I think it's the best version to start a prototype with.

Shoot-through is respectable with only one large spike over a 10 cycle period at full power/20khz. At anything less than full power, it is virtually eliminated.

Coupling caps can easily be added for anyone with a DC offset problem on their source, without affecting performance worth mentioning. It simulates well with a DC offset of even 100mV, but it will offset the output as well.

I'm not concerning myself with common mode voltages induced in the feedback paths, I think that's best left to a proper layout.

I'll take a few days to complete another another range of simulations to fully prove the design, and I'll include the shoot through graphs as well.

From what I see thus far, however, THD results seem a bit worse than those first posted here, but not much, and are still somewhat impressive.

Regards,
Chris
 
Class D Starting Point?

I followed the thread "Development of a Class D Starting Point".. because I am interested, and this is the outcome.:bigeyes:

I hoped for a gentle introduction of the principles, and maybe a simple circuit to experiment with.

If this is the starting point, where will it end?:eek:
 
Re: Class D Starting Point?

johnnyx said:
I followed the thread "Development of a Class D Starting Point".. because I am interested, and this is the outcome.:bigeyes:

I hoped for a gentle introduction of the principles, and maybe a simple circuit to experiment with.

If this is the starting point, where will it end?:eek:

Well the outcome of the starting point thread... is in the starting point thread :)

That's why I created a new one for this version, which isn't starting point. I offer all schematics and everything though, It might be a good idea to try simulating playing around with it... good way to learn a bit, I think.

If you dig up one of the schematics of the half bridge given in the other thread I think you'll find what you're after, I've prototyped an even simpler stripped doing version of one of those, just the basic comparator and half bridge stage. Works great for what it is, but it's not something I want to listen to if you know what I mean, it was never meant to be, considering it worked at all though I don't think it's anything to be afraid of. It has annoyed the neighbors for extended periods of time though.

The next circuit I'll be posting here, isn't really any more complicated than the first one. I'm unable to comment on the others. ;)

I recommend you read the patent for the introduction to the principles. It's linked in the first thread someplace.

If you're interested in trying an ultra simple stripped down version I'll get you the one I used, as I never posted it.

Regards
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.