| Workhorse |
Hi Folks,
I have successfully implemented this elevated rail bootstraping technique to prevent failure mode in IR2110 gate driver during low frequency heavy clipping conditions...
Your comments please....
Kanwar |
|
|
| phase_accurate |
R U sure you have drawn it right ?
One possible solution can be found on IR's website (some charge pump using a 555).
Another one would be the use of a simple current-source from your elevated rail. You wouldn't need much current supplied by this (in the range of 350 uA IIRC).
Regards
Charles |
|
|
| BWRX |
| Can't you just use the positive rail, an emitter follower, and a larger capacitance? |
|
|
| Workhorse |
| Mine isnot a current source, its a voltage source type of 15V floating and is referenced to the source....I have tested the amp with DC at input also and the output easily delivers it without any damage in continuous load.... |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
Can't you just use the positive rail instead of the "elevated" rail? |
When you want to drive low frequency signals, the charge in bootstrap caps decays and the Mosfet enters linear conduction mode and would get damage,but using this elevated rail of VCC+15V the cap always remains charged, even if DC signal is used at the input of amp[provided DC coupling]... |
|
|
| BWRX |
| quote: | Originally posted by Workhorse
using this elevated rail of VCC+15V the cap always remains charged |
I realized that after I posted, and edited my previous post :cannotbe:
Your scheme makes sense but if you don't plan on amplifying DC wouldn't it be easier to just increase the value of the bootstrap cap? |
|
|
| Workhorse |
| With this technique the amp can act as continuous DC voltage source.... |
|
|
| phase_accurate |
The base voltage must be higher than the emitter voltage to turn on the transistor. So I think you have drawn the circuit differently than you actually implemented.
| quote: | | Can't you just use the positive rail, an emitter follower, and a larger capacitance? |
This wouldn't help since it wouldn't allow sustained turn-on of the upper MOSFET which is what is intended here. The idea is to allow driving the amp heavily into clipping without causing latchup or worse.
Not that heavy clipping should be a usual mode of operation of any amp - but it is always good practice to design things such that they survive any type of misuse.
Regards
Charles |
|
|
| Workhorse |
| 0-100% duty cycle could be realized,no smoking of amp during continuous clipping of low frequency signals at output.... |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
The base voltage must be higher than the emitter voltage to turn on the transistor. So I think you have drawn the circuit differently than you actually implemented.
Regards
Charles |
Yes,
Actually
The elevated Rail voltage is 25V
The Zener voltage is 16V and thus the floating bootstrapping voltage is around 15V[including losses]....perfect to drive the mosfet during heavy clipping |
|
|
| phase_accurate |
OK I C.
Regards
Charles |
|
|
| Workhorse |
Now I am happy driving any type of abusive clipped program continuously...
Do you like this technique Charles?:) |
|
|
| BWRX |
| quote: | Originally posted by phase_accurate
This wouldn't help since it wouldn't allow sustained turn-on of the upper MOSFET which is what is intended here. The idea is to allow driving the amp heavily into clipping without causing latchup or worse. |
| quote: | Originally posted by Workhorse
0-100% duty cycle could be realized,no smoking of amp during continuous clipping of low frequency signals at output.... |
DC and heavy clipping are two different conditions. You don't think you could get away with using a much larger bootstrap capcitance for the heavy clipping condition? Obviously it won't work for DC... |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
DC and heavy clipping are two different conditions. You don't think you could get away with using a much larger bootstrap capcitance for the heavy clipping condition? Obviously it won't work for DC... |
I have tried 220uFD cap also, but encountered a failure during driving subsonic clipping signals around 18Hz.....
So I developed this technique and now quite happy with the results so far |
|
|
| phase_accurate |
| quote: | | Do like this technique Charles? |
Yes I do. A disadvantage might be the third rail. But this voltage can be generated on board as well if necessary. You know I am fond of topologies that are easy to use like the UcD modules for instance where you only need ground and two rail voltages. This is of coarse a non-issue for complete amps.
I was once thinking about some charge-pump technique to charge the bootstrap capacitor.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
Yes I do. A disadvantage might be the third rail. But this voltage can be generated on board as well if necessary. You know I am fond of topologies that are easy to use like the UcD modules for instance where you only need ground and two rail voltages. This is of coarse a non-issue for complete amps.
I was once thinking about some charge-pump technique to charge the bootstrap capacitor.
Regards
Charles |
Thanks!
I think generating third rail from additional windings of regular transformer[or from on board DC-DC converter] isnot difficult, when the amp under consideration is designed for power output in Kilowatt levels and with pro world abusive conditions survival...
Atlast I feel relief from the clipping headaches...they are gone forever...
regards,
Kanwar |
|
|
| BWRX |
| quote: | Originally posted by Workhorse
I have tried 220uFD cap also, but encountered a failure during driving subsonic clipping signals around 18Hz... |
That's a huge bootstrap cap! Hard to believe it couldn't hold the voltage up long enough. What voltage is it being used at and how much current does the cap have to supply? |
|
|
| phase_accurate |
I just guess but maybe recharging this thingie posed a problem.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
That's a huge bootstrap cap! Hard to believe it couldn't hold the voltage up long enough. What voltage is it being used at and how much current does the cap have to supply? |
I onced used it with subwoofer class-D amp with SW 75Khz, the Bootstrap voltage was 15V, cap rated at 25V, The IR2110 was used to drive 2 pairs of IRFP460[as their gate charge was high thats why the cap wasnot charging enough during massive 12dB signal overdrive conditions]....
The input signal was 18Hz and the rail voltage was +-130V[open load] , load was 2 X 18" woofers @4ohm total |
|
|
| phase_accurate |
The input capacitance should have nothing to do with the current drawn while being on. The static current is only drawn by the driver circuit.
The turn-on is of course drawing charge from the bootstrap cap but this should be the same per switching event wheter the amp is overdriven or not. So your trouble might have been caused by something else than the high input capacitance.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
The input capacitance should have nothing to do with the current drawn while being on. The static current is only drawn by the driver circuit.
The turn-on is of course drawing charge from the bootstrap cap but this should be the same per switching event wheter the amp is overdriven or not. So your trouble might have been caused by something else than the high input capacitance.
Regards
Charles |
But when I applied my new technique to this amp , the failure was gone....;) |
|
|
| BWRX |
You would hope so since you now have a constant voltage supply :)
From your description it seems like the bootstrap cap was never fully charging (as Charles stated) or you had a high amount of leakage current causing it to discharge too quickly. |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
You would hope so since you now have a constant voltage supply :)
From your description it seems like the bootstrap cap was never fully charging (as Charles stated) or you had a high amount of leakage current causing it to discharge too quickly. |
Could you more ellaborate on leakage current.....phenomena
But now it became much more robust with stiffened constant voltage source;) |
|
|
| BWRX |
| There's certainly some leakage current due to stray capacitance but it shouldn't be much if the layout was good. The other possibility could be the driver chip itself, as Charles mentioned. |
|
|
| alexclaire |
Hi Workhorse,
Happy to see that someone had the same problem i encountered with IR2113...
I tried different values for bootstrap capacitors and in hard clipping conditions output stage blew up...I think it was because of the bootstrap capacitor not charging enough to make the upper fet stay in saturation...
Can you tell more about your design ?
values you tried, fets you use...
Thanks |
|
|
| BWRX |
| quote: | Originally posted by alexclaire
Happy to see that someone had the same problem i encountered with IR2113... |
You shouldn't be happy that he was having problems ;)
| quote: | | I tried different values for bootstrap capacitors and in hard clipping conditions output stage blew up...I think it was because of the bootstrap capacitor not charging enough to make the upper fet stay in saturation... |
Were you using the recommended application circuit where the bootstrap cap is charged by a Vcc of about 15V?
If so, why don't you guys try charging the bootstrap cap from the positive supply rail with a series resistor. Place a zener in parallel with the cap to set what voltage you want it to charge up to. |
|
|
| Workhorse |
| quote: | Originally posted by alexclaire
Hi Workhorse,
Happy to see that someone had the same problem i encountered with IR2113...
I tried different values for bootstrap capacitors and in hard clipping conditions output stage blew up...I think it was because of the bootstrap capacitor not charging enough to make the upper fet stay in saturation...
Can you tell more about your design ?
values you tried, fets you use...
Thanks |
The EF BJT is MJE340, Diodes are UF4007, Zener is 16V 1W type, and the resistor, thats the home work for you, as its rating is dependant on rail voltage and idle current as well...
I have used IRFP250/260/360/460, APT20M18LVR, APT30M35BVR etc.....and some IXYS stuff also...
| quote: | Originally posted by BWRX
You shouldn't be happy that he was having problems ;)
Were you using the recommended application circuit where the bootstrap cap is charged by a Vcc of about 15V?
If so, why don't you guys try charging the bootstrap cap from the positive supply rail with a series resistor. Place a zener in parallel with the cap to set what voltage you want it to charge up to. |
Thats known as PRE-Charging, but it doesnot help during Prolonged clipping events as the output gets strucks to the upper rail and then there is negligible current flow from that resistor to the cap....
But here in my technique, even when the voltage at output is struck to rail, the gate to source pulse voltage amplitude is still 15V constant, no matter how long its time interval is...Thats the figure of Merit for my technique....
Kanwar |
|
|
| odnaizutra |
hi Workhouse,
what are the advantages of using your method instead or ir ones using 555? |
|
|
| BWRX |
| quote: | Originally posted by Workhorse
Thats known as PRE-Charging, but it doesnot help during Prolonged clipping events as the output gets strucks to the upper rail and then there is negligible current flow from that resistor to the cap.... |
If the capacitor can keep the FET fully on during the prolonged clipping event then it will help. Using the positive rail will allow you to fully charge the cap up to a higher voltage so that it will keep the FET fully on for a longer period of time. How much voltage can be used depends on what the chip and FET can handle.
Using a separate constant voltage supply will of course work under any conditions. The trade off is just added complexity. |
|
|
| Workhorse |
| quote: | Originally posted by odnaizutra
hi Workhouse,
what are the advantages of using your method instead or ir ones using 555? |
1.It saves PCB area and helps in minimizing parasitics
2.It has very much less component count [6 components] than NE555 charge pump type approach has
3.It is very very simple and much better
4.It is much robust and concrete solution |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
The trade off is just added complexity. |
A handful of 6 components is an Added complexity
A handful of 6 components which guarantee your amp would be safe in delivering DC, any type of Clipping output without any type of abnormal behaviour is an added complexity....
A handful of 6 components which could safeguard your FETs and Gate Driver Chip is an added complexity...
Just a handful of 6 components!:bawling: |
|
|
| BWRX |
| quote: | Originally posted by Workhorse
Just a handful of 6 components!:bawling: |
and an extra transformer/winding... |
|
|
| Bender.ru |
| quote: | Originally posted by BWRX
and an extra transformer/winding... |
so we can use it as floating voltage supply for hi side driver, and there is no need in 6 extra components :D |
|
|
| phase_accurate |
| quote: | | Using the positive rail will allow you to fully charge the cap up to a higher voltage so that it will keep the FET fully on for a longer period of time. How much voltage can be used depends on what the chip and FET can handle. |
I don't see much of an advantage doing this.
| quote: | | and an extra transformer/winding... |
As already mentioned this wouldn't be much of an issue for a finished industrial product.
If one is selling modules then he better generates all auxiliary voltages on board since ease of use is always a good selling point. Lars Clausen's earlier modules needed an auxiliary voltage that he now "generates" on board -and I don't think that sales dropped when he made that decision.
As already mentioned you can also supply this current by a current-source from an auxiliary rail. It is not much current that is needed - less than 500 uA. For an amp with +- 100 Volt rails this would mean an incrase in losses of less than 50 Milliwatts. Not that much for an amp that could supply 600 Watts approx into 8 Ohms. The advantage of the current-source method would be less parasitic capacitance than any other method.
Regards
Charles |
|
|
| IVX |
IMO, useless handful
1) i don't remember if 22-47uF of bootstrap, would be not enough for any realistic Qg and audio content (20hz 100% clipping even). DC = emergency, and amp must be shutdown instantly.
2) ir2110 already have under voltage protection (UVP<8V) for both outputs, to preventing poor Rds_on. |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
The advantage of the current-source method would be less parasitic capacitance than any other method.
Regards
Charles |
But if current demand exceeds, there must be excess voltage drop accross the cap...isn't it[provided very large Qg of Mosfets]
I think voltage source will help in maintaining the voltage level...much easily....
regards,
Kanwar |
|
|
| Workhorse |
| quote: | Originally posted by IVX
IMO, useless handful
1) i don't remember if 22-47uF of bootstrap, would be not enough for any realistic Qg and audio content (20hz 100% clipping even). DC = emergency, and amp must be shutdown instantly.
2) ir2110 already have under voltage protection (UVP<8V) for both outputs, to preventing poor Rds_on. |
If it was that much reliable than reputed manufacturers like Crown, Powersoft, Ecler wouldnot use a "dedicated" floating high side supply rather than a simple bootstrap cap in their class-d amps....If they follow your type of opinion, then its Kaboom! |
|
|
| Eva |
Actually 2.2uF is far enough if things are done in a clever way.
It's just a matter of adding some logic to force the high side to be turned off for, say 1us, if it has been on for the last 100us or so...
The low side capacitor has to be substantially bigger, though, say 22uF, to avoid draining it too much during the charging pulses.
Actually, the under-voltage lockout does a great job and the real problem is usually the low side capacitor being temporarily drained too much and causing other circuits to malfunction. If the high side capacitor is 220uF charged to 7V (UVLO threshold) and the low side capacitor is 22uF charged to 15V, the low side capacitor will be discharge to 7V almost immediately after the UVLO is triggered... This leads to a funny situation if more stuff is being powered with those 15V...
The too low UVLO thesholds of those chips may be a problem too, because Rds-on becomes too high before UVLO is triggered. However, the "watchdog timer" logic solves the problem.
|
|
|
| Workhorse |
| quote: | Originally posted by Eva
Actually 2.2uF is far enough if things are done in a clever way.
It's just a matter of adding some logic to force the high side to be turned off for, say 1us, if it has been on for the last 100us or so...
The low side capacitor has to be substantially bigger, though, say 22uF, to avoid draining it too much during the charging pulses. |
My earlier setup has 47uF for Bootstrap and 470uF for low side cap, but upon severe 6 to 11dB overdrive conditions it got smoked...but with my new technique it never smoked up again...mine is a low cost yet robust and reliable way to completely eliminate the problem.
|
|
|
| darkfenriz |
Kanwar
Why not just add a voltage source to a source of upper mosfet i.e. parallel to normal bootstrap cap? |
|
|
| Workhorse |
| quote: | Originally posted by darkfenriz
Kanwar
Why not just add a voltage source to a source of upper mosfet i.e. parallel to normal bootstrap cap? |
You mean to say completely isolated floating 15V referenced to source, it could be done very easily, but the floating supply then would become a stable source of great EMI radiation inferno!!! |
|
|
| darkfenriz |
| Why, even decoupled? |
|
|
| phase_accurate |
I was once thinking of the watchdog thingie as well but I like the approach of constantly powering the high-side driver better.
Regards
Charles |
|
|
| IVX |
| Indeed 2.2u will almost enough to heavy clipping on the very low frequency, though ir2110's typical -UVP treshold is 8.3V@75C (9.5V worst case), and Vbs static idle current 350uA (worst case), hence 12V[Vdd, no more if fastest switching is needed]-.6V[diode's drop]-50mV[to single charge 100-200nC old hat MOSFET]-9.5V=1.85V; 2.2uF*1.85V/350uA=11.6mS i.e. 43hz 100% clipping ok, so would be the best choice >4.7uF yet. |
|
|
| Workhorse |
| quote: | Originally posted by darkfenriz
Why, even decoupled? |
You cannot decouple this w.r.t. ground, you have to use a common mode choke to decouple the floating voltage source....
| quote: | Originally posted by phase_accurate
I was once thinking of the watchdog thingie as well but I like the approach of constantly powering the high-side driver better.
Regards
Charles |
Thanks for the vote of confidence!Sir
The watchdog, The chargepump, The fully isolated floating supply each one is having there own merits and demerits, but my elevated high side bootstraping technique is most simple to do and easy to achieve with the required results so far.....
Regards,
Kanwar |
|
|
| Bender.ru |
| quote: | Originally posted by Workhorse
but my elevated high side bootstraping technique is most simple to do and easy to achieve with the required results so far.....
|
Really? you think so? Maybe in DC convertors with const duty cycle =1 your technique adaptable, but in audio range amp it has more disadvanatage, than benefits...imho ;)
I suppose the case, that transform your amp to smoke generator differ from version with hi side driver problem. |
|
|
| phase_accurate |
| quote: | | Really? you think so? Maybe in DC convertors with const duty cycle =1 your technique adaptable, but in audio range amp it has more disadvanatage, than benefits...imho |
I don't know how you come to this conclusion since it should be vice versa IMO.
Regards
Charles |
|
|
| IVX |
| Guys, what about you talking? I've just check it in the MC, it's completely wrong. See attachment, where my mistake? :cannotbe: |
|
|
| phase_accurate |
Yu forgot the resistor between the base and the auxiliary rail.
Regards
Charles |
|
|
| IVX |
| haha, yeah, i see now, ok it's work, what about 3 diodes around follower, superfluous ones? |
|
|
| phase_accurate |
IMO D6 might not be necessary but D5 will protect the transistor from reversing on transients.
Regards
Charles |
|
|
| Eva |
| I think that the pass transistor is going to have a really hard time handling the high dV/dt switching transients without conducting more current than what it is expected to. Also, the bias resistor has to provide enough current to bias the zener and the pass transistor with only 10V across but it also has to be able to whitstand the full 200V supply across it too, which seems like a big compromise... |
|
|
| phase_accurate |
| quote: | | I think that the pass transistor is going to have a really hard time handling the high dV/dt switching transients without conducting more current than what it is expected to. |
That's why I would prefer a current source that is supplying this 350 uA.
Regards
Charles |
|
|
| IVX |
| BTW, Q1 will dissipate 4.35W (to-220+heatsink?) at 200khz 200+25V 50% duty cycle with irfb4227. |
|
|
| IVX |
| Charles, rather 30mA current source, but not 350uA one, with the same dissipation at Q1 >3W. |
|
|
| phase_accurate |
No, no, no and no again ! You would only have to supply the quiescent current of the driver. The pulse is still supplied by a small bootstrap cap as usual. The thingie is just there to prevent undervoltage lockout for any arbitrary clipping situation.
Regards
Charles |
|
|
| Eva |
Does anybody know a current source capable of whitstanding 5V/ns slopes whithout too much miller effect?
:xeye: |
|
|
| IVX |
| quote: | Originally posted by phase_accurate
No, no, no and no again ! You would only have to supply the quiescent current of the driver. The pulse is still supplied by a small bootstrap cap as usual. The thingie is just there to prevent undervoltage lockout for any arbitrary clipping situation.
Regards
Charles |
The pulse is still supplied by a small bootstrap cap +DIODE as usual? |
|
|
| fredos |
Why not use standard bootstrap and the circuit of WorkHorse only when output are near voltage rail? More efficient and less dissipative! Or like I use in some serie of amplifier, a bootstrap circuit with a 555 charge pump just to witsthand clipping condition...More efficient! 4 watts of lose is too much in a class d amplifier. The trick is only to find a way to hold voltage of hight side at clipping...
Fredos |
|
|
| phase_accurate |
| quote: | | Does anybody know a current source capable of whitstanding 5V/ns slopes whithout too much miller effect? |
Common-base circuits do not suffer from Miller-effect. Apart from that a little Miller effect doesn't matter at all. The current supplied by this auxiliary current source would only be needed during a static condition.
Tonio:
Nice idea indeed but it might also suffer from draining under certain conditions.
Regards
Charles |
|
|
| IVX |
| quote: | Originally posted by fredos
Why not use standard bootstrap and the circuit of WorkHorse only when output are near voltage rail? More efficient and less dissipative! Or like I use in some serie of amplifier, a bootstrap circuit with a 555 charge pump just to witsthand clipping condition...More efficient! 4 watts of lose is too much in a class d amplifier. The trick is only to find a way to hold voltage of hight side at clipping...
Fredos |
Hi Fredos. Did you had any troubles during clipping, if standard bootstrap technique used (i mean ir2*** etc with UVP)? |
|
|
| IVX |
| I guess, that some troubles are possible, but due to other reason, and maybe at startup rather. For instance, one of most successful and reliable class D designs (like UcD) use 100uf bootstrap cap only, however idle current (ON state) ~4 times more then 350uA, so for ir2110 it's equal to 22uF. |
|
|
| Pierre |
Are you sure it is 100uF? It sounds like a lot to me.
For IR2110 I have had very good results with 1uF. |
|
|
| IVX |
| UcD180 -100uF, UcD400 -220uF, but UcD drivers aren't CMOS like the ir2110, and their static ON_state consumption >1mA. |
|
|
| fredos |
Just remember that a 20Hz hardly clipped signal need a hold time of 25 ms....That's means that you need large capacitance to hold it at a minimum of 10-12V to avoid that mosfet go into linear mode... Add to this idle current of driver itself, miller effect draining, temperature coefficient of component and leakage of capacitance....Bigger is better, but on other hand, it need more charging current, and in same way, provide more discharging current, lower the life of this capacitor. I use a 470uF, aluminium military type, with a 555 charge pump when output (common of mosfet) reach rail supply. So far, that's the only way I have found to got good result. Auxiliary supply is a good way too, but increase EMI if a single supply close to mosfet is not used for each top side.
Fredos |
|
|
| phase_accurate |
For those who want to use a large cap I already made the following suggestion once (don't want to search):
Use a small cap as usual as main bootstrap cap like 100 nF for instance. Then use an "emergency cap" of larger size of which current is drained during long on-periods of the high-side driver. In order to make this current available quickly when needed without soaking the lower-driver's supply (as soon as the lower-side FET is switched on again) you use a Schottky diode and a parallel resistor between the two caps.
Regards
Charles |
|
|
| Pierre |
My experience with bootstrap cap is that, with higher values (I originally used 10uF), there was an annoying "tick" when powering up and down. With 1uF, it has almost disappeared.
Clipping behaviour is correct, on the other hand, even for low freqs.
Charles, some more details about your last suggestion?.... Thanks! |
|
|
| IVX |
| UcD use 33kOhm pullup, for precharging bootstrap cap, and zener to clamp it at the 12V. I don't know too, why need the dedicated source or other bruteforce solution if standard circuit ok.. BTW, ir2110 and similar drivers, just can't provide 2-3A, even pulsed, long time at about 80-100C, so they need P-BJT follower at least, that will often safer for the ir2110 if the mosfet blew up. |
|
|
| phase_accurate |
| quote: | | Charles, some more details about your last suggestion?.... Thanks! |
Below you can see what I am talking of. The components in the box are the additional ones.
Regards
Charles |
|
|
| lumanauw |
Hi, Charles,
What happens if I use only single bootstrapp capacitor Elko type, 100uF/50V without any 100nF in parrarell? Is this not good for operation of 400-500khz? |
|
|
| BWRX |
| quote: | Originally posted by IVX
UcD180 -100uF |
My older UcD180 (11/04/2005) uses a 220uF 16V cap.
| quote: | Originally posted by IVX
I don't know too, why need the dedicated source or other bruteforce solution if standard circuit ok.. |
That's my thinking too. The brute-force method (a constant high side supply) should and obviously does allow the chip to work properly at high levels of low frequency clipping and even DC, but a larger bootstrap cap should allow proper operation even at high levels of low frequency clipping. That makes it sound like there was some other problem that made the bootstrap cap discharge too quickly or did not allow it to fully charge. |
|
|
| TOINO |
Hmmm... What about a secondary on the output filter coil? a few turns would suffice to give 10 or 12V for supply the gate driver.
It is also possible to derivate a kind of overload protection from that voltage…
:rolleyes: |
|
|
| lumanauw |
Hi, Toino,
That's a clever design. I've tried such a cap multiplier for high-side, but coming from SMPS transformer (f=30khz), at that time it doesn't work, I think because the switching frequency of the classD high side is much higher than the SMPS (400khz compared to SMPS' 30khz) |
|
|
| Bender.ru |
| quote: | Originally posted by TOINO
Hmmm... What about a secondary on the output filter coil? a few turns would suffice to give 10 or 12V for supply the gate driver.
It is also possible to derivate a kind of overload protection from that voltage…
:rolleyes: |
Hmmm... what about hi side clip? Your solution suitable for 0<duty cycle<1 only, but don't work at duty cycle=1 :rolleyes:. |
|
|
| phase_accurate |
| quote: | | Hmmm... what about hi side clip? |
The watchdog would be better than that IMO.
| quote: | | What happens if I use only single bootstrapp capacitor Elko type, 100uF/50V without any 100nF in parrarell? Is this not good for operation of 400-500khz? |
I would definitely use a cap with low ESR close to the driver.
Regards
Charles |
|
|
| Bender.ru |
| quote: | Originally posted by phase_accurate
The watchdog would be better than that IMO.
|
other solution is forced limitation of PWM duty cycle
(0.05..0.95), that's allow avoid clip and also there is no need in huge military cap (470u ;) ) |
|
|
| Workhorse |
| Forced limit on duty cycle , but there are still some drawbacks associated with it as it adds additional "Pulse Edges" which might interfere with the switching cycles ....This technique is used in Crest amp LT1200..... |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
That makes it sound like there was some other problem that made the bootstrap cap discharge too quickly or did not allow it to fully charge. |
It could be another problem[parasitics, leakege current], but again brute force solves it....It has some drawbacks too as its subjected to high dissipation, but thats not so much of the concern when output power level is in kilowatts rather than watts... |
|
|
| Workhorse |
| quote: | Originally posted by Eva
Does anybody know a current source capable of whitstanding 5V/ns slopes whithout too much miller effect?
:xeye: |
Yes I know one , its used in ECLER DT4800..It uses a cascoded current source for stable reference along with additional voltage source as well....The SW is 250kHz...The pass Transistor is ofcourse a Mosfet rather than bjt... |
|
|
| Workhorse |
| quote: | Originally posted by BWRX
The brute-force method (a constant high side supply) should and obviously does allow the chip to work properly at high levels of low frequency clipping and even DC |
The Brute Force!:) sometimes good sometimes bad :D |
|
|
| Iyremenko |
| Why someone didnot suggested, the use of auxillary full floating high side supply of 15VDC with reference to source of high side mosfet. Any problem associated with, i think is common mode filtering issues only. |
|
|
| Workhorse |
| quote: | Originally posted by Iyremenko
Why someone didnot suggested, the use of auxillary full floating high side supply of 15VDC with reference to source of high side mosfet. Any problem associated with, i think is common mode filtering issues only. |
That would be also great solution, but in that case it needs to be placed near the high side mosfet to get lower EMI |
|
|
| phase_accurate |
Another simple solution that would not even have to be isolated would be a flyback converter consisting of a coil that is connected to the low-side driver supply with its first leg. The second leg is periodically pulsed low to the lower rail by a small FET and it's freewheeling current is "discharged" to the high-side bootstrap capacitor via a small and fast diode.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
Another simple solution that would not even have to be isolated would be a flyback converter consisting of a coil that is connected to the low-side driver supply with its first leg. The second leg is periodically pulsed low to the lower rail by a small FET and it's freewheeling current is "discharged" to the high-side bootstrap capacitor via a small and fast diode.
Regards
Charles |
How about using the clock[assume a forced clock design] itself in providing pulse for driving isolated transformer for floating supply need, just drive the clock into set of buffers powered from preamp +/-15V supplies and their outputs attached to the floating tranformer primary and you have 2 sets of isolated voltages in handy!!! |
|
|
| phase_accurate |
I would not even dare to use anything else than the amp's clock to do this ! :cool:
An isolated supply has the disadvantage that it relies on isolation. This is not difficult to achieve from a DC point-of-view but rather tricky regarding the capacitive coupling between primary and secondary.
A non-isolated flyback circuit would not be susceptible to such problems.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
I would not even dare to use anything else than the amp's clock to do this ! :cool:
Regards
Charles |
I was not referring to 250Khz clock for amp, but i was referring to the lower frequency sync clock generated from the same clock driver.......;)
Can you more ellaborate on capacitive coupling between pri-sec disadvantage in class-D......
Kanwar |
|
|
| phase_accurate |
Well - the coupling capacity of your transformer is acting like a capacitor between the switching output signal and ground. This might result in quite high current spikes.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
Well - the coupling capacity of your transformer is acting like a capacitor between the switching output signal and ground. This might result in quite high current spikes.
Regards
Charles |
High current spikes.........but where?? |
|
|
| phase_accurate |
| quote: | | High current spikes.........but where?? |
Most probably into the signal ground of the control stages of your amp - since you want to feed this auxiliary PSU from the low voltage rails.
Regards
Charles |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
Most probably into the signal ground of the control stages of your amp - since you want to feed this auxiliary PSU from the low voltage rails.
Regards
Charles |
Ok....thats very good point ........
So you are against contamination of signal ground:D
thanx |
|
|
| phase_accurate |
I am against contamination of ANY grounds !!!
Regards
Charles |
|
|
| Workhorse |
| I think we can also power the flyback from the main rails and then use isolating DC-DC converter to get 2 X +15V for hi & lo sides respectively......without taking the ground in to play!! |
|
|
| Workhorse |
| quote: | Originally posted by phase_accurate
Well - the coupling capacity of your transformer is acting like a capacitor between the switching output signal and ground. This might result in quite high current spikes.
Regards
Charles |
I have to disagree on it now.......after testing one of the reputed Class-D amps in real life with isolated gate driver voltages obtained from +15V preamp supplies......there is no trace of current spikes anywhere.....
The amp which was tested is from Powersoft |
|
|
|
