| alexclaire |
Hi everyone,
D amp Proto is running fine !
Dead time adjustments made...sound is great !
here are some pics of the proto:
post filter output voltage ripple 4Vpp@ 250kHz
no shoot through anymore ;-)
thanks Fredos |
|
|
| alexclaire |
| another one of the proto : |
|
|
| alexclaire |
and i forgot
output DC offset : 18mV |
|
|
| alexclaire |
Hi everyone
just a little question for Fredos
with no load, the amp has sometimes trouble to start...
maybe because of the feedback loop : loudspeaker is part of the output filter ;-)
I think it's normal (amp with no load isn't a usual way to do !!) |
|
|
| zilog |
| How do you take care of the gate voltage supply for the high side FET? Maybe your schematic requires grounding through the load in order to transfer charge to this supply. |
|
|
| alexclaire |
Hi Zilog,
Supply of the upper fet is made by bootstrap diode and capacitor through the load you're right ! |
|
|
| luka |
Hi
Isn't 4Vpp at 250kHz a bit too much? Is this because of series resistance of filter cap in output filter? |
|
|
| luka |
Hi
And what is your supply DC voltage when the amp is idle on what when at full power? |
|
|
| alexclaire |
Hi luka,
filter corner frequency is around 25kHz, and triangle wave is 250kHz
so one decade away...
LC filter is second order so 40dB/decade attenuation
So fundamental amplitude of output pre filter square wave is divided by 100 (40 dB)
so it should be around 2Vpp (instead of 4vpp).
DC bus voltage is +/-82V at idle
and +/- 74V at max power 4 ohms load |
|
|
| luka |
Hi alexclaire
- Can you also tell me why is D8 (mur120) placed between pin2 and 5 of IR
- Are pin3 and 4 of 4070 floating?
And could you point me to thread where I could read about 555 current limiter cos I didn't find anything but just mentioning it |
|
|
| alexclaire |
Hi Luka,
D8 (MUR120) is just here to prtotect the mosfet driver (IR2113)
it acts when Vs tends to become lower than -Vcc (in case of capacitive load)
yes pin 3 and 4 of 4070 are not connected... |
|
|
| luka |
Hi alexclaire
Are those 10k resistors cross Gate Source really needed? |
|
|
| alexclaire |
hi luka
don't think are necessary
gonna change output stage:
gonna put diodes in serie with mosfet and one anti parallel to prevent bodydiode current flow
mosfet die when high dv/dt or di/dt because body diodes are too slow |
|
|
| luka |
Hi alexclaire
Do you really think diods in series is necessary, coz you would need big diods, put them on heatsink. If it without for you now and without destruction of fets I won't them in. |
|
|
| Workhorse |
| quote: | Originally posted by alexclaire
like this |
Crest CD3000 schematics distributed by me are now seems to be everywhere ....;) |
|
|
| Workhorse |
Check This Out
www.microsemi.com/datasheets/APTM20AM10STG-Rev3.pdf
Series Schottky Silicon Carbide Diode + Anti Parallel Fast Recovery Diode + NTC+ 200V 100A Mosfet in HalfBridge all in one isolated package and also low EMI radiation
makes this device an ideal for high power Class-D Audio amps...I am going to order these for sample evaluation....
Kanwar |
|
|
| sixtek |
Hi,
| quote: | | makes this device an ideal for high power Class-D Audio amps |
Hmmm... I wouldn't say ideal.
Check capacitances (input capacitance: 14nF) and the diodes characteristics...
Not really suitable output device for class-d amp.
I'd rather use simple high power mosfets. There are many good devices on the market.
The choice is yours!
All the best,
PS.: I'm really courius about the amplifiers that manufactured by your company, I prefer studying hi-power technology. Shall you give us the web-address of your audio company? |
|
|
| luka |
Hi alexclaire
Near end with designing ;)
 
Any comments? |
|
|
| lindomar |
Hi LUKA.
Is new version of ARES Proteus 7.1? |
|
|
| luka |
Hi
It is Ultiboard from National Instruments |
|
|
| alexclaire |
Hi everyone,
gonna make changes with D AMP output stage but need some information about antiparralel free wheeling diode...
Gonna put schottky common cathode in serie with mosfet drain (diode is STPS3045CW) in order to prevent current from flowing through the fet body diode (too slow) and force it to go through the outside diode (antiparrallel).
I'm wondering which recovery time I should use (depends on deadtime i guess) trr=40ns is good ? or do i need less (25-28 ns) ?
With 24ns deadtime (RCD circuit before IR 2113 Low and high inputs R=47ohms C=470pF) when I try to increase output signal to reach saturation output stage burns before saturation...
Should i lower dead time as output stage didn't burn with low deadtimes ?
I guess body diode remains reverse conductive as the other fet is turned on...
someone got an idea ? |
|
|
| phase_accurate |
What is your switching slew rate ? Be careful not to turn the "inactive" FET on via the reverse capacitance when trying to switch too fast.
Regards
Charles |
|
|
| luka |
Hi alexclaire
But changes to output aren't really necessary, right? It works even without antiparralel free wheeling diode..., fets are just heating more, right? |
|
|
| Workhorse |
Hi Charles,
Can we replace the series Schottky's with this....
The centre node is a combination of 2 series nH inductors which ADD a little oFFset to body diode of mosfet, thus enabling the Fast recovery diodes to perform efficiently...
Kanwar |
|
|
| phase_accurate |
Hi Kanwar
This wouldn't prevent the turn-on of the body diode. But it might help a little to reduce the reverse recovery current peak. But my gut-feeling tells me that I don't like it somehow.
Regards
Charles |
|
|
| Workhorse |
So there's no replacemant for robust schottkys ....
Any another Idea in your mind... |
|
|
| Bender.ru |
| Hi all. Here is my light version of Fredos Damp. |
|
|
| Bender.ru |
Here is my schematics & PCB (Sprint Layout 4.0).
______________
Have anice day. |
|
|
| luka |
Hi
Damn this looks really great, how the sound? |
|
|
| Bender.ru |
Hi Luka.
All i can say, clear treble, really deep bass :-).
I like it much more than, tda8924 or gainclon from 729x series.
Some spec:
THD(1kHz 1W 6.5Ohm) 0.04%,
THD(1kHz 60W 6.5Ohm) 0.12% ,
THD(1kHz 140W 6.5Ohm 0.18%
AES-17 24k lowpass filter, SBLive, Spectrolab
DC offset - 15mv |
|
|
| phase_accurate |
Very nice ! Did you test into what range of loads it is stable ?
Regards
Charles |
|
|
| IVX |
| Hi Bender, i guess you got pretty good switching, though dead time isn't 15nS yet. Your circuit have about 14db loop gain, i.e. THD reduced just by factor 5 vs open loop, for instance Hypex's UcD loop gain is about 30db or factor 30, so 1/6 of the your THD figures seems very close to UcD400 performance, hence precision of the switching very close for both too. If not so much big idle consumption, then try to short the gate's diodes etc, BTW, PSRR is pretty poor, due to low loop gain, it's could be improved (with the same loop gain) by H-bridge or in the UcD manner (res from VCC and res from VDD to the inverting comparator input). After all you can change the feedback loop, and get a lot of loop gain with WOW THD figures. (hint: ask Charles) :) |
|
|
| fredos |
Low gain reduce distortion, that's why I put it so low in my design. PSRR still very good due to this low gain. You can improve lot more THD figure if you reduce amplitude of triangle wave, but check for stabilty when you go with load and without load. Lower the amplitude of triangle wave (open loop gain), better performance you will get, but more subject to instability.
Fredos
PS higer voltage=more stability!
Soft clipping can be attained by matching gain of first stage (TL072) to lower gain of power section.... |
|
|
| Bender.ru |
Hi Charles, i run amp with 2Ohm, 4Ohm acoustic and 6.5 Ohm resistive load. It works stable, problems appear without zobbel, also i've got trouble, when try to reduce trianlge amplitude close to 0.1-0.3 V (in my proto triangle =0.5V).
Hi Ivan, nice to see you. Yea, DT=35ns, idle consumption ~4.5W.
Reducing DT from 45nS to 25nS provide THD reduction in 1.5 times and increase idle consumption from 4.5W to 5.5-6W. Main intend of amp is subw application, that's why i stopped at 35nS DT.
Next device will be H bridge with 100kHz carrier.
But it really interest to get lower THD and PSRR, i hope Charles help me with FB :rolleyes:.
Hi Fredos, you are right about triangle amplitude, schematics become instable. PHP28NQ15 DS voltage limits power supply voltage. |
|
|
| IVX |
| Bender, THD of the such carrier based amps, very depends from the propagation delay (but selfoscillating amps completely other story), so faster comparator in your case must be preferable. If you need only subwoofer amp, 100khz will quite enough, however output impedance rather too high for sub_amp (damping factor 20@2ohm= just a shame), and bass will noticeable boominess. Did you measure the output impedance? |
|
|
| Bender.ru |
| IVX, fast comparator is good, but seems to me that cheap & popular 319 enough for 100kHz career. Didn't measured output impedance yet, i'll do it in near future. My ears said, that damping factor much better than 20@2Ohm, i hope they are right :clown:. |
|
|
| IVX |
| Yeah, it's my mistake, at 1khz DF= 20..30, but not at the bass! :) |
|
|
| Bender.ru |
:D, i'll measure DF & post result as soon as i can.
ps:i designed new PCB with similar output stage, but PW modulator using UcD principle. Maybe it show better result, than my first one. |
|
|
| luka |
Hi
How do you measure output impedance at some freq.? |
|
|
| IVX |
| It's usual test (AC source->Rload->Routput). Where the problem? |
|
|
| luka |
Hi
And how do you measure Routput? I can't just measure it with ohm-meter. Do you need to know Iout, voltage drop cross inductor + fet? |
|
|
| Bender.ru |
| Hi luka, here is simple schematic, describing measurement of output impedance. Note: sinewave gen must provide output currents up to ~1..3A. |
|
|
| Bender.ru |
| Amp schematics with fixed errors. |
|
|
| luka |
Hi alexclaire, Fredos
Is it ok if I put 330pF or 470pF instead of 390pF (C4)? |
|
|
| alexclaire |
Hi Luka,
My proto works with 330pF so don't worry about it ;-)
I'm looking for a safe way to drive output N channel Fets...
I've burnt several IR2113 and I'd like to find another solution...
I am trying fredos's model 1200 way to drive P and N channel fets but limited in power because of P channel poor specs...
If you have any idea
I think my problem is with bootstrap cap...
Help is needed !
thanks |
|
|
| fredos |
Fairchild FQA36P15, free sample on web site, matched with IRFP250 or IRFP260 work well, at a maximum of +/-70V. You will have to play a bit with gate resistor to match dead time to these device. Another way is to double IRFP9240 with same gate resistor and a IRFP260. I have do that for special order for a customer that use it bridged into 3 ohms for bass instrument with succes.
With this configuration, you will take advantage of adaptive dead time! You will like it!
Fredos |
|
|
| alexclaire |
Hi Fredos !
thanks for your answer !
you mean double P channel with only one N channel ?
your model 1200 was biased with +- 80 v rails wasn't it ?
a bit much for safe operation with irfp240 and 9240 no ?
what do you think of IR211X drivers ? |
|
|
| fredos |
+/-80V with switching power supply is not a probleme at this level. IR211X serie of driver are very sensitive to everything. I only use it in switching power supply for my HVI serie Amplifier. There are not enought rugged for class d amplifier. Yes I means 2 P channel with 1 N channel. Unusual, but work very well. At +/- 70V you can use with no prob the Fairchild I suggest you. At 36A iD, these device work well to 2 ohms per channel! Order it as sample from Fairchild! These series of amplifier are on the road for more than 10 years and still rock many bar and band! They only use cheap parts and are very reliable. This is what we can call a proven old design! :-). We will maybe release a new budget serie with this old design with these new Fairchild parts, under maybe 400$ each amplifier for 600 watts per channel. Like I told you, they use only easy avaible and cheap parts, that was not the case 10 years ago! Just follow exactly schematics for power amp section and you will not be dissapointed.
Fredos |
|
|
| alexclaire |
Hi Fredos,
thanks for your answer !
Gonna try with such device....
sound with your model 1200 front end and ir2113 was great i'm not disapointed at all ;-)
very excited to try your design very smart indeed the way you apply negative feedback to comp input ! seen nowhere else ! |
|
|
| fredos |
I still use this technic in my lattest serie of amplifier! Sound great, work well and really stable! I did'nt find a better way from now!
fredos |
|
|
| IVX |
| quote: | Originally posted by fredos
I still use this technic in my lattest serie of amplifier! Sound great, work well and really stable! I did'nt find a better way from now!
fredos |
Did you measure THD/Freq plot? I think it's would be pretty flat, though claims about zero IMD fully marketing BS, obviously. BTW, the trick to have more loop gain with the same stability, does exist, i guess, that i see it.;) |
|
|
| fredos |
To add more gain, you need error amplifier that add delay in amplifier transient respond. Using comparator as error amplifier/comparator combo avoid this, at the expense of lower gain. That's why I claim that IMD is near zero! You can put a 20 Khz squared wave at the input and you will get a squared amplified wave at the output, with no overshoot! And yes, you definitively need lo pass filter at the input to avoid frequency over audio spectrum....Basicly, UCD work in the same maneer (no error amplifier), but in free running mode....Just dont use clock in my circuit and you will got same result!
Fredos |
|
|
| IVX |
| So, do you've THD/Freq plot yet? Error amp, to say more exactly -integrator, it's too simple way, though 60-80db of loop gain (feedback after filter) at 1khz would be very impressive for carrier based PWM. I mean light tweak by RC network modification only. |
|
|
| phase_accurate |
The trick to make loop gain higher could be achieved by making the triangular smaller in amplitude. This could go ahaed with increased triangular linearity. Since a higher triangle voltage is desirable in practice - because it makes the modulation process less susceptible to EMC problems - we would have to make a tradeoff unfortunately.
Regards
Charles
Edit: And yes, stability could be improved by adding a resistor somewhere ! ;) |
|
|
| fredos |
We use a old calibrated HP analogue THD meter. We cannot under 0.01% of THD with accuracy. All of our amplifier are under 0.1% of THD at full power, and under 0.01% from 10W to full power, from 30-20 000 hz. We always got a dip at 6.3Khz, due to the feedback compensation. Under 10 watts, THD cannot be measured due to the residual noise at the output. Even with a 22Khz low pass filter, meter are unable to performe. I Know that for purist, these spec are not impressive, but my market is (hight) power amplification. And as Alexclaire said, he seem impressed by the overall sound quality of the amplifier, and all of our customer too.
Your right too that with lower triangle wave amplitude higger gain could be achieved, but under certain level, amplifier go out of stability. Low amplitude can be attained by placing physicaly on PCB the integrator capacitor and resistor really close to the input pin of comparator with good ground plane around it. Lower the value of the resistor and higger the value of the capacitor provide enought margin again EMI and stability.
Fredos |
|
|
| IVX |
| quote: | Originally posted by fredos
We use a old calibrated HP analogue THD meter. We cannot under 0.01% of THD with accuracy. All of our amplifier are under 0.1% of THD at full power, and under 0.01% from 10W to full power, from 30-20 000 hz. We always got a dip at 6.3Khz, due to the feedback compensation. Under 10 watts, THD cannot be measured due to the residual noise at the output. Even with a 22Khz low pass filter, meter are unable to performe. I Know that for purist, these spec are not impressive, but my market is (hight) power amplification. And as Alexclaire said, he seem impressed by the overall sound quality of the amplifier, and all of our customer too.
Your right too that with lower triangle wave amplitude higger gain could be achieved, but under certain level, amplifier go out of stability. Low amplitude can be attained by placing physicaly on PCB the integrator capacitor and resistor really close to the input pin of comparator with good ground plane around it. Lower the value of the resistor and higger the value of the capacitor provide enought margin again EMI and stability.
Fredos |
nice if the sound great, by the way, not every professional class D amp can sound even decent (many people tell me, that Powersoft, Crest etc class D purchased for light weight only, but never for the sound), next for the decent THD spec, you need exactly those 6db of loop gain, that you lose in the your current design, with the same part count. |
|
|
| fredos |
We got better result with X17 gain in the power stage!
Fredos |
|
|
| raintalk |
| quote: | Originally posted by IVX
nice if the sound great, by the way, not every professional class D amp can sound even decent (many people tell me, that Powersoft, Crest etc class D purchased for light weight only, but never for the sound), next for the decent THD spec, you need exactly those 6db of loop gain, that you lose in the your current design, with the same part count. |
People I talk to say just the opposite. Lightweight is an added bonus. |
|
|
| alexclaire |
Hi Fredos,
:bawling: proto has just burned...
don't know why
output stage blown (IRFP 240 9240)
with no signal and 9.4 ohms load
output ripple was 2vpp @ 250khz but with high voltage overshoot at every switching
that's a problem of deadtime i think
the values were taken from your model 1200
Vcc is +/-60V
just changed rgate from 4.7 to 10
and REG+ and REG- resistor from 510 to 680 (53ma per amp isn't it ?)
help
amp failed wen i applied 1kHz sinewave low amplitude to input |
|
|
| luka |
Hi
I hope you will put let say 100 ohm or more resistor each in path from supply to amp, for + and -. This will help if something will go wrong next time |
|
|
| alexclaire |
hi luka !
i used to do that but not this time...
I changed the output stage P N channel fets
but something is wrong with dead time adjustement...
shoothrough burned the 2 fets... |
|
|
| luka |
Hi
Burned those fets??? That must have been large shoothrough. Hope you will start to use those resistors again until you are sure that it works |
|
|
| alexclaire |
the 2 fets are dead with a hole in the center of the package...
i suppose this is due to shoothrough
what could it be with a little input signal ?
i need some help to fix this |
|
|
| luka |
Hi
Fredos will know this best, but it could have oscillated |
|
|
| alexclaire |
| here is a pic before "the accident" ;-) |
|
|
| luka |
Hi
This is how my looks like now
 |
|
|
| alexclaire |
Hi luka,
you made the previous version with IR2113 driver ? |
|
|
| luka |
Hi
Yes is did, but took one step forward from yours |
|
|
| alexclaire |
what are transistors tip50 for ?
voltage reg ? |
|
|
| luka |
Hi
Well you had supply voltages coming from other board, that i've eliminated by making those tip50 voltage regs. So you only need main supply voltage and thats it. |
|
|
| fredos |
Have you exactly follow schematics? I have built over 500 of these amp and never get any probleme with them....Maybe the 10 ohms gate resistor, but I dont think....This should just add more dead time...How was your square wave before it blow? Overshoot? Used 0.47uF MKT between mosfet Source? Zener level shifter was there? You can play a bit with the emiter resistor of the level shifter to play with dead time (R13 and 14 I think...Dont have schematic with me... By the way, I use Litz wire for output coil... Better result everywhere...Maybe this?
Fredos |
|
|
| fredos |
Hum....Did see your MKT decoupling cap on your picture! C1 and C2 on schematics! That's your mistake!
Fredos |
|
|
| alexclaire |
Hi Fredos,
Thanks for your answer but what caps are you talking about ?
C1 and C2 aren't decoupling caps...
MKT aren't suitable for decoupling with such frequencies ?
Got one 1uF cap in the input path (blue one) (high pass filter with 10Hz -3dB frequency ) and the other is part of the triangle wave integrator (yellow one 4.7nF) but I used these ones with the previous proto wit no problem...triangle wave is clean...
And for the MKT at the input, at worst it has an impact on the sound quality but not on amp reliability ?
Could you tell more about the dead time adjustement ?
would it be good to replace R13 and R14 with variable resistor to set DT easily ?
Dead time control module is here to provide more dead time when output is low isn'it ? by increasing emitter current no ?
At the output of your module at idle you have a clean sinewave @250Khz with ~2Vpp ? do you see switching spikes on it ?
before it blew i saw such 20V spikes at each switching...
I forgot i didn't have 0.47uf between sources, the orange one you see is part of the zobel network (10 ohms + 220nF)
The output filter cap isn't good ? 0.47uF MKP 600V ? |
|
|
| alexclaire |
Hi Fredos,
i've just disconnected output stage and i'm making some measurements on the modulator and level shifter.
for the upper stage of the level shifter i measure 5,93V across R5 680ohms is it enough tu turn the upper fet on ?
ie half time 60v half time 54,07V
on the lower side i measure 4,93V across R6 ?!
ie half time -60V half time -55,07V
maybe fets burned because they were in linear mode and not saturated...
i've tested adaptative dead time:
with 0v at the output : 50% square wave across R5,R6 (680)
with 6V at the output : 0% square wave across R5,R6 (680)
don't understand how it could work like this fredos
could you tell me more ?
thanks for all ;-)
|
|
|
| alexclaire |
Sorry Fredos
just understood the way it works
i've injected DC at the output but i dind'nt pay to attention to the feedback loop so duty was changing !
when i disconnect feedback dead time variation appears...so no problem
upper side 5.9V across R5 to drive fets with 0V output
upper side 7.8V across R5 to drive fets with 16V output
lower side 5V across R6 to drive fets with 0V output
lower side 7V across R6 to drive fets with -16V output
enough to turn on fets ? |
|
|
| alexclaire |
| here is a drawing...Vout=+4V |
|
|
| luka |
Hi
He means the ones going from +vcc to -vcc and are 0.47uF (C1,C2 on his schematics)
PS: 3 to 4 volts are needed for turning on fet, so with 7v fet is fully opend |
|
|
| luka |
Hi Fredos and Alexclaire
I have finished building amp from alexclaire schematic, and it is working great. Thank you Fredos for making and posting the schematics of the amp, it is the best amp ever, and thank you alexclaire for those pictures and schematics that you send me.
You two are the best:worship:
I will post some pic tomorrow, have to get some sleep, been working on amp from beginning to the end for 7h non-stop, but well worth it:dead: |
|
|
| fredos |
Yes, MKT decoupling cap are between mosfet source, it is very important for ringing. I play on dead time with gate voltage, less voltage means lower dv/dt on gate=less dead time. So dead time are lower at low level to reduce distortion and higger at high power to got better effiency. Scope the drain of mosfet and adjust dead time with level shifter emiter resistor to got minimum overshoot. When you got minimum overshoot, you are at optimum dead time setting at idle. Over this, dead time controler will take care of hight power dead time. you can put 2 trim pot to adjust it at R13 and R14. Monitor too the rail current. Minimum current is just after optimum deadtime, so you can trim amplifier idle loss versus optimal deadtime. Normally, each rail should be adjusted for 20-30ma at idle, so about 3 watts of loss at idle. Check too if ringing is not generated by your coil by simply unplug it and compare with and without. If your interested, I think that I have few board left of these serie, but it was first production without deadtime control, but is really easy to add it by rework.
Wish that is helping you!
Fredos |
|
|
| alexclaire |
Hi Fredos,
Thanks for all ! gonna check all this !
I'm interested in trying your boards or the ones you were talking about in an other thread ;) 200W module that a "famous" amp maker brand left you. Tell me your conditions...
Bye.have a nice day.
Gonna have some time to experiment : i'm on holidays and 27°C :) |
|
|
| luka |
Hi
Here are the pic
  |
|
|
| alexclaire |
i Luka !
could you post some results ? some measurements ,
or pics of signals... |
|
|
| zkaiser |
| Yes and please try 20kHz sine wave from sound card as input and tell me do you hear some artefacts on speaker. Or if you are able to hear something, that means that your amp is in some resonace and not working propperly for 20Hz-20kHz audio range. |
|
|
| luka |
Hi
I don't think there is anything like that since this is crystal controlled amp. I will use function generator for waveform. |
|
|
| luka |
Hi
Here are some pic
This one is from output sine wave 250kHz, 0.5v/div, spikes because of poor decopling cap cross +/-Vcc, maybe aslo because of too big inductor, don't know the value of it.
And this one is from output also near clipping. 10v/div, with input on 1v/div (upper one). Amps supply is from car smps at +/-30v
Have no problems going into max clipping, so that output is squere wave
What do you think? |
|
|
| luka |
Hi
And these are the synchronizing clocks, 125kHz for smps and 250kHz for other amps
 |
|
|
| alexclaire |
Hi luka !
Nice results...
The spikes you see on 250khz sine wave are due to dead time i think...
try to make it longer and you'll see these spikes disappear...
50Vpp at clipping that's it ?
so 78Wrms under 4 ohms load...
I got no problem till 60Vp so 450WRMS under 4 ohms...
at 60Vp got some ringing at the top of the sine wave.
i pushed it to 144Vpp so 650WRMS
do you implement overload protection onboard ? |
|
|
| luka |
Hi
Not jet. Tomorrow. When I said near clipping I did mean that one side of signal was near, not both (problem in generator), so there cant be normal clipping on top and bottom of signal. I should say that at this point I didn't had load connected. Reason for that is that I don't have right equipment (supply for amp). So for now I am limited to +/-30v one, but still with some mod in feedback I will raise it to +/-50v.
Dead time? I will look at this. But I don't think there in any throughshoots, because there is low consumption.
I have looked at this just now and I see 3A spikes on series resistors. This is indicating that there are throughshoots. How much should I increase resistors? They are 100 now. |
|
|
| Bender.ru |
Hi Luka, here is some of my pics:
LC output, 250kHz 0.5V/div, dead time = 35nS (there is no spikes, as you can see ;) ) |
|
|
| Bender.ru |
| Next one, totem pole, 20v/div: |
|
|
| Bender.ru |
| Clipping. 2mS/div, 20V/div, 0.47uF bootstrap capacitor. |
|
|
| phase_accurate |
That is a very clean and exemplary behaviour !
Regards
Charles |
|
|
| luka |
Hi
Yes bender that is something.
This is pic of current through 0.1 resistor when scope is on 0.1v/div. smaller is current or upper fet, bigger is lower
this was taken with dead time trimers set to 470 both. It seems that I can't close lower fet fast enough. Am I right? |
|
|
| phase_accurate |
I am probably wrong but this looks like parasitics to me.
Regards
Charles |
|
|
| luka |
Hi
Could be that problems are coming from fact that I use smps to power it. Amp and smps aren't synchronized. |
|
|
| Bender.ru |
| No. Imho problem in pcb topology & DT, try non-inductive 0.1Ohm resistor. |
|
|
| luka |
Hi
PCB could be a problem, you can see it on my webpage,DT even a lot higher don't cure this, but it does lower current a lot. I do think I have inductive resistor, but I don't know if I can get non-inductive ones. |
|
|
|
