UCD 25 watts to 1200 watts using 2 mosfets

Well, that really I can not explain all that well. The only thing is that the way it is connected to the comparator is faster than how it connects normally. But this has the disadvantage of a higher percentage of imbalance and greater amplification still counting. All this was explained by the author here. Amplificador HighEnd Clase D de 25W a 1250Wrms sólo con 2 mosfets N - Página 55 Would translate as much text is here. But that explains the connection of the comparator. And is not unique. Saludos!
 
Well, that really I can not explain all that well. The only thing is that the way it is connected to the comparator is faster than how it connects normally. But this has the disadvantage of a higher percentage of imbalance and greater amplification still counting. All this was explained by the author here. Amplificador HighEnd Clase D de 25W a 1250Wrms sólo con 2 mosfets N - Página 55 Would translate as much text is here. But that explains the connection of the comparator. And is not unique. Saludos!

This post is still about offset correction.
To make myself clear I made schemtics. So this is the current connaction of the comparator:
An externally hosted image should be here but it was not working when we last tested it.

As you can see feedback goes to In+, signal is taken from Gnd, and to pin 7 +3V is given. In this connection according to the datasheet, the LM311 has about 1us delay. This delay is higher than any other part in the schematic.

This is the connenction I am talking about:
An externally hosted image should be here but it was not working when we last tested it.

As you can see feedback goes to In-, signal is taken from pin7 (OUT), pin 1 (gnd) is taken to gnd... This is the normal connection of the comparator, according to the datashett it has 200ns delay.
The shorter dealy could help is take the switching frequency higher. In ther original schematic it is about 120-150kHz, which is very low, and causes lot of dissipiation in the output coil. I managed to take it up to 200-250kHz at lower supply voltages, but to take higher we need to shirten the propagation delay.
High propagation delay also limits the maximal modulation limits. So the amp cannot be driven to its supply rails at the output, at about 80-90% modulation it starts to produce modulation error, producing THD. If the prop delay is shorter, thaen the modulation limits can be pushed much higher, so you would need a little smaller rails to produce high power at certain THD.

Tekko was doing experiments, said he pushed the switching freq to 450kHz by this normal connection of the comparator, he published his PCB plan and schematic, but without component values.

I could not be the only one, who is interested in this, this modification can improve the amps perfomance much more.
As I said previosuly DC offset is about a zero problem. If it exists, there are hunderd easy ways to take it away, its a pity job.
 
Hi lorylaci,

I tried this in LT-Spice:
Leave the feedback on the positive input of the comparator.
Use a pullup at the output (7), as you have done, take the signal from there.
Connect pin 1 and 4 of the comparator to -3V. (you have pin 1 on gnd).
Swap HIN and LIN of the IR2110.
At least in the simulation it works.
 
Hi lorylaci,

I tried this in LT-Spice:
Leave the feedback on the positive input of the comparator.
Use a pullup at the output (7), as you have done, take the signal from there.
Connect pin 1 and 4 of the comparator to -3V. (you have pin 1 on gnd).
Swap HIN and LIN of the IR2110.
At least in the simulation it works.

Yesterday my firend told me the main problem: "you have pin 1 on gnd, so your comp output swings between 0 and 3V, not enoguh to turn on the transistors next to". I put pin 1 to -3V. Now everything works. With some setups my switching frequency reached 400kHz. But now I use the following schematic as a test:
An externally hosted image should be here but it was not working when we last tested it.

The switching freq is at 320kHz (3,1us periodic time). (the same feedback with other connecntiion was at 250kHz)
 
1200 Watts

hi
greetings any information on this amp?
 

Attachments

  • t1200n.jpg
    t1200n.jpg
    117.1 KB · Views: 1,697
kit

DIGITCLASS ™:.
1. Set Kitts miniXD Power Amplifier.
- Self assembly kit Kitts.
- Grading of high-quality equipment all.
- Complete set of equipment including heat sink plate bolts, etc. Mika.
Price: 850 Baht

2. Preset to print available miniXD Power Amplifier.
- Power Supply + /-45V to + /-90V power sub + BIAS 12V.
- W is up to 700W @ 4 Ohms, 1200W @ 2 Ohms.
- Of a printed and completed all the test board.
Price: 990 Baht



3. Sheets printed miniXD Power Amplifier.
- Use the print sheet EPOXY FR-4 is already green.
Price: 35 Baht




4. T157-2 core wire thousands of 22uH.
Price: 200 Baht HAS ANYONE TRIED THIS KIT
THANKING YOU
andrew lebon
 
DIGITCLASS CLASS-D POWER AMPLIFIERS MODULE T2400H.

T2400H SPECIFICATIONS:.
• 400W @ 8 ohms (+ /-90V).
• 750W @ 4 ohms (+ /-90V).
• 1400W @ 2 ohms (+ /-90V).
• 2200W@1.33 ohms (+ /-90V).

• 800W @ 8 ohms (+ /-125V).
• 1500W @ 4 ohms (+ /-125V).
• 2400W @ 2 ohms (+ /-125V).

• Input Impedance: 47k ohms unbalanced.
• Frequency Response: 20 Hz - 20 kHz.
• Supply Voltage: + /-45V to + /-125V.
• BIAS Voltage: 12V to 14V.
• Dimensions (W x H x D): 55 x 45 x 150 mm.
Price: 850.00 Baht


DIGITCLASS ™:.
DIGITCLASS CLASS-D POWER AMPLIFIERS MODULE HyperXD.

HyperXD SPECIFICATIONS:.
• 400W @ 8 ohms.
• 750W @ 4 ohms.
• 1400W @ 2 ohms.
• 2200W@1.33 ohms.
• Minimun load 1 ohm.
• Input Impedance: 100k ohms unbalanced.
• Frequency Response: 20 Hz - 20 kHz.
• Supply Voltage: + /-45V to + /-95V.
• BIAS Voltage: 12V to 14V.
• Dimensions (W x H x D): 55 x 50 x 200 mm.
Price: 1,290.00 Baht

woots:.
Released when offline is very interesting.

Old.
 
Of course you can use anything, if there is no other solution, but:
This FET has a very high gate charge. You must use SPadi's idea of the extra transistor helping turn of. It also has very high body diode charge. So this amp won't be so efficient. Care must be needed to avoid cross-conduction too.

I'd reccomend full-bridge design, with lower voltage. At full-bridge you can go up 90-95V, there is no bus pumping (less safety margin is neeed, now you use 250V fet for +-90V).
 
@ Lorylaci

I bought on the flea market toroid transformer 1500VA 2x62V and 3000VA 2x50V.

First one I wanna use in half bridge but obviously only good transistor for that voltage for 4ohm load is IRFB4227

And second transformer I wanna use in full bridge (IRFB4227) 4 ohm load with good heatsink for my brother disco club but I have to find little toroid core with low Al (for overload protection about what you told me)

Laci, what about your schematic with changes on LM311. Is this good? What if I change 330p to 100-150p and output filter to 20uH with 1uF + IRFB4227 :)?

P.s both transformers I bought for 20e from gipsy (probably stolen) :)
 
@ Lorylaci

I bought on the flea market toroid transformer 1500VA 2x62V and 3000VA 2x50V.

First one I wanna use in half bridge but obviously only good transistor for that voltage for 4ohm load is IRFB4227

And second transformer I wanna use in full bridge (IRFB4227) 4 ohm load with good heatsink for my brother disco club but I have to find little toroid core with low Al (for overload protection about what you told me)

Laci, what about your schematic with changes on LM311. Is this good? What if I change 330p to 100-150p and output filter to 20uH with 1uF + IRFB4227 :)?

P.s both transformers I bought for 20e from gipsy (probably stolen) :)

Yeah, you can get good copper-cables or heavy transformers from gipsies any day.

I haven't tried the current transformer overcurrent protection, If you want to try, I can send you the schematic, but I don't want to post it here, until I haven't tested it (some people says it would newer work).

The normal connection of LM311 decreases propagation delay of the whole circuit about 10-20%, so switching frequency and modulation limit increases. If you have a good IRFB4227 or some other premium FET, you can take switching freq easiyly up to 300kHz, and then 20uH coil is enough (less wire, less losses...), so you can get really high powers, from small size. Since modulation limit increases, with the same supply voltages you can get higher power at the same THD level.

See the schematic I posted.

With so much power, you must use really big caps to avoid bus pumping at half bridge. If you can get 250V fets (IRFB4229) then it can be much safer at 2x90V supply.
 
I have 8 pcs of Simens Elko 100V 30000uF each, but they are realy big because they are old 20 years ( from old german radio station).

In fact all the same except of part with LM311. I can rewrite the old PCB. What about 10uF cap on pin 6 of ir2110 and UF4004, same quation for gate diodes 1N5817/19. Can all be the same except of part with LM311.

On monday I wanna build this amp with changes because tomorrow in the afternoon pcb will be completed
 
Here there have been some improvements over the DC offset at the output. Engineer Eduardo has provided an explanation of why the commitment to design in the use of the comparator and its consequences. Translated yet one can understand.


"""PS: To Tacatomón: What servo to correct the offset I thought several times, but it's just best not to. The designs that use the matching problems are discrete components at the low signal.
Since we are using an integrated comparator, the offset should not be necessary to compensate, because embedded within the comparator transistors are exactly alike.
But the design of this UCD has a small "failure" ... : For the offset of a comparator input as the LM311 transistor as small as possible, which is the impedance of the inverting input of the comparator to ground should be equal to the impedance is non-inverting input to ground . That is to offset the current comparator inputs generate the same voltage drop across each of the input resistors connected to these inputs.
But in this design UCD, one of the inputs are approximately 820 ohms, and the other 47kohms. That is what creates an imbalance that many are having - and which, moreover, is amplified by the gain of the amplifier.
When I design, I knew about that problem ... but could not lower the input impedance of the audio signal to 820 ohms, it is usually too low for what are the vast majority of signal sources, overloading.
Nor could he raise the impedance resistive divider feedback, because if the impedance is too high, it starts to pick up noise.
So I took that might have some offset to the output (because it does not really affect performance too.)
But after what was discussed here a long time, and something adrian2008 said, I realized that it was possible to obtain the matching of impedances that are the feet of the comparator, which solves the problem of offset it, without losing either, impedance audio input, or increase the noise picked up by the feedback ... Sometimes a bit hard to notice these things ... Especially since they consider as an ideal comparator, the ultimate strength that advised to adjust the offset that added magnetrón27 not serve any function, but it's true in this world of actual comparators ... And when magnetron account of the noise problem that appears, then I remembered that resistance and, as stated tacatomón, aerial act secure .. And for that, the cure is a capacitor in parallel, because the comparator offset current does not flow the same, but act short-circuit for noise and interference (ALWAYS, the average noise voltage is 0, but could wear resistance of thermal power generators!)

Hopefully this will resolve this issue definitively offset!
"""
 
Last edited:
Has anyone had an overheat in the bjt transistors stage?

I built the amplifier to 400W@4ohms,,power supply= +-60vcd , but the transistor stage(shift level) gets so hot that it cant be touched. It sounded good but then Q1 and Q4 transistor blowed up.:flame:

When I replaced these transistor the overheat still. So, I have changed the values of R3=1K instead of 120 ohms. R9,R8=18K instead of 2.2k.

The heat goes down. When I turned on it again it sounded at least some seconds then Q4 transistor blows up again. The ir2110 was dead. :/ :RIP:



What could be happening ?

P.D.Muerto = dead :dead:


I had troubles with zener-transistor regulator , it blowed up too , when i turned the volume to max. Now it´s solved with an external power supply.
 

Attachments

  • transmuertos.PNG
    transmuertos.PNG
    41.3 KB · Views: 1,493
  • DSC01889_1024x768.jpg
    DSC01889_1024x768.jpg
    94.1 KB · Views: 1,426
  • DSC01891_1024x768.jpg
    DSC01891_1024x768.jpg
    113.8 KB · Views: 1,373
Last edited:
Has anyone had an overheat in the bjt transistors stage?

I built the amplifier to 400W@4ohms,,power supply= +-60vcd , but the transistor stage(shift level) gets so hot that it cant be touched. It sounded good but then Q1 and Q4 transistor blowed up.:flame:

When I replaced these transistor the overheat still. So, I have changed the values of R3=1K instead of 120 ohms. R9,R8=18K instead of 2.2k.

The heat goes down. When I turned on it again it sounded at least some seconds then Q4 transistor blows up again. The ir2110 was dead. :/ :RIP:



What could be happening ?

P.D.Muerto = dead :dead:


I had troubles with zener-transistor regulator , it blowed up too , when i turned the volume to max. Now it´s solved with an external power supply.

Are you sure they are 2N5401? Did you checked the footprint? Other manufacturers can use other footprints, check the legs with a transistor tester.
Since your trnasisturs burned, the HIN and LIN inptus of IR2110 got high voltage, so this is why it burnt. The Q1-Q4 should disspate less than half watt, they should be lukewarm only.
You should not change the leverl shift resistors, otherwise the delay of level shift will be very slow, the given values are quite good.
Your reglutaor for Vn+12 burnt? Are the FETs okay?
Are you sure you have 2x60V DC?
If you chosed your regulator resistors well, they should be warm, but they should take.
If you turn up volume, bus pumping due to bass can raise the power rails.
There must be lots of stuff you missed, or misconnected, or did not measured or calcualted well. Check all the values after.
How much capacitors do you use? What was the load (pure resistance or speaker) What was the supply?

The original design itself is not reccomended over 200W, because there is no enough decoupling at the supply rails. There should be some elkos near to extend peak current, and cermaic capaitors under them to extend high frequency current.
 
OMG!
Now i see you used IRFP460 !!!!
I'm sorry to say that, but you must have been really stupid, first read after, then try.
(Or you will learn it on your own fault. - as now)

The IRFP460 Fets are very old 500V Fets, they have very high gate charge. They intended to use at n*10kHz SMPSs, not in Class-D amp.

You should use the lowest (but in safety range) voltage rated fet. You must cehck for gate charge, body diode charge and rdson. If body diode charge high, you must use diode trick. If gate charge high, you must use extra transistor to help turn off, or totem pole. IF rdson is high, but any other is low, you wil have high conduction losses. IRFP460 is in ALL WAY a bad choice

Try some IRFP250N or IRF640N if you can't get any good, modern fets. And first read some...

I'm sure you had cross-conduction, and I think everything was conducting everywhere...
 
Yes, I did same mestake with IRFP260N and this PCB without elkos near output fets and high voltages for rails. 2N5401 blow up after several minutes and IR2110 was deat. After few days when I bought IRFP250N everuthing was all right.

Laci, PCB are finished :). I have components and good IRFB4227. With very short wires (something like jumpers) I'll change conections pins of LM311. Cap of 330p to 100p and output filter to 20uH. I'll post the photos.
 
Ty for repply.

My first try was with the irfp250 , but i haven´t more of them. I know irfp460 its slow but i havent any problem with the mosfet stage.

I use mpsa92 instead of 2n5401.I checked the footprint and it´s ok.

(irfp250)I had one prototype working a long time , but when i did the test to get max power the power supply blows up again.(12v supply Zener-transistor ,Opened Zener).I measure atleast 150W@4ohms (4ohms speaker and 6ohms pure resistor.)


I thought that i got bus pumping effect when i was trying bass tracks. But now trying with 1Khz signal it happened again.
I measured all supplies.
10,000uF by rail

I use bypass diode , mur120
Snubber 100pf-10ohm , I did not see spikes in the output stages.

Also the amp its oscillating at 150Khz

I read all thread , spanish and english threads.

I change the Rg resistor 27ohm to 22ohms.Whit the 12v external supply to the ir2110 the signals looks better less noise and oscillations.

The pics below are with irfp250.

i´ll upload more captures soon.
 

Attachments

  • señal gate mosfet.png
    señal gate mosfet.png
    47.1 KB · Views: 1,342
  • Fosc_Salida.png
    Fosc_Salida.png
    19.2 KB · Views: 1,248
  • salida filtro.png
    salida filtro.png
    16.8 KB · Views: 393
  • salida.png
    salida.png
    15.9 KB · Views: 419
Last edited:
Put Elkos near by ouptut fets. For that PCB put them under board, 470uF will be good. Change cap of 330p to 100p, and swiching freq will be around 260KHz. Output filter from 30uH to 20uH and cap from 1.5uF to 1uF.

My amp works good with that changes, even on 2ohm load and IRFP240 on +-50V