| Whortless |
Hi to every body,
I started to desing a class-D amp(enclosed shematic). I would likte to make it stereo for full range at least 100W/4ohms. I used modified self oscillating iraud amp1. Also I choose HIP4081 instead of IR2110/2113. As I showed in the shematic, now I am using IRFP250 as output mosfets. I know they are slow, has high Qg and other worse parameters but they are working in my bread bord without heating. I am now using only one channel oscillating about 220KHz. I also tested IR2113 with same shematic but output mosfets warm up. Again I tested IR2113 with IRF540(ST) and saw high side ise still warming up. Therefore I think my last desing is ok for only 12V. For final desing, I think I need level shift and symetrical PS for input opamp and output mosfets. Now all of them running on the supply voltage of mosfet driver.
I need some help on calculating closed loop gain of this desing.
Any suggestions will help much from you experienced DIY'ers.
Regards |
|
|
| Whortless |
| Are not there someone to give me some idea? |
|
|
| luka |
Hi
I can't help you with loop gain, but I have a question.| quote: | | I also tested IR2113 with same shematic but output mosfets warm up. Again I tested IR2113 with IRF540(ST) and saw high side ise still warming up. |
What was your capacitance for upper, HI side drive? |
|
|
| Whortless |
Hi Luka,
If you are asking about C8 it is 680nf. But tried 220nf and get it working without heat of high side mosfet and UVL protection of HIP4081. And now I am using 100nf+1uf(electorlitic). No problem with HIP4081 there is no heat on both mosfet. I also replaced R105 4.7K to 1K and now oscillating frequency is 330KHz and better sound.
Regards |
|
|
| luka |
Hi
You used 680nf with IR? That is too small. Use 22uF or more, when you will go in hard clipping you will need it. I use 100uF elco and 22uF tantal in parallel, with no heating in upper mosfet |
|
|
| Whortless |
OK
I will take this suggestion to care. But IR2113 failed in my most desings. I did some SMPS circuits with 2113 also one of my friend, who has a scop(oscilloscope), did a fulbridge SMPS with IR2113. We saw there with scop the H side mosfet is not turned off properly although there is a dead time on input stage of 2113. Therefore I used HIP4081 because, it has delay time adjusment for H and L side. You can have a look to data sheet and also application note of HIP4081 on www.intersil.com . The delay resistors with 30K there is a dealay lets say about 30-40ns or so.
Kind regards |
|
|
| luka |
Hi
Could you post or send me schema of that smps or amp (if you did it with IR), I'll take a look of what you used.
I have both smps and ClassD amp running with IR and that is with no problems. Didn't destroyed not even one IR.
Hip are great but limited by max voltage. If you will be making one with HIP you should look here first.
But you made this one
Didn't it work? From that picture I made my, which look pritty much the same |
|
|
| Whortless |
Hi luka,
As I saw you know what I did before. OK the scop photo belong to driver transformer version of my SMPS. But, both of them still working. The problem is IR version of my SMPS heating a bit up. Also scop result of IR version like a sinus. I enclosed the shematic of my smps' IR version. And also many thanks to you for the HIP link. |
|
|
| Whortless |
My final desing enclosed. The input section of CH-B is same with CH-A. I put there a level translator but, i think i should put a reverse polarity 1N4148 to B-E legs of Q105. +/-Vss is about 35 to 38V so the final total power supply 70 to 76V.
Ps: My circuit is not a full bridge. But ,stereo signals will be 180 degree phase angle with additional input buffers. That will give me the chance of briging two channells. |
|
|
| luka |
Hi
I can tell you right now that 0.47uF is not enough, use much high C and I would use irfp450 reather then 460,rest will check later. |
|
|
| Whortless |
Thanks Luka,
I will test boostrap cap with higher values like 47uf low ESR+100nf mylar. Also I have a working version with IRFP450 but with same boostrap cap.
Regards |
|
|
| luka |
I forgot to say that you should use something like 47uf low ESR +100nf mylar for lower drive too. U have to have this for upper and lower side drive because you have to charge G fast from ~0 to Vcc (12V). If this is not done as it should be, fet will go in so called linear area=:flame:
I would also use 7812 instead of 15. It is enough and heat will be lower. I would remove R109 and R104 (not needed).
Other than that I don't see anything wrong for now. Did you test it? Like current limit, is feedback stable? |
|
|
| Whortless |
Hi luka,
I was originally removed R104 and R109. I know that IR2113 has shimt trigerred inputs so it is correcting the worst square waves. But, I do not have scop for now. But I can say it is working better and also FB was stable when I created it.
Regards |
|
|
| luka |
Hi
For this kind of things like smps's spoce is a MUST |
|
|
| Whortless |
For this kind of things like smps's spoce is a MUST [/B][/QUOTE]
I know but, do not have. I can borrow it from a friend. But not now only weekends. Do not forget please I am a Chemist in fact :).
Kind regards |
|
|
| Whortless |
Hey Luka and others,
My friend could not bring the Scop this weekend because of his wery hard works on 1200VA three phase smps. Any way, I did a test pcb for HIP4081 and worked well, both channells :). See the enclosed photos. I also remove 4081 and then attached IR2113 with some wring by inreasing the boostrap C to 47uf+100n. Now, This pcb is not my final desing, I did this only for test to see the layout behaviours and for final pcb. Althoug 2113 above the PCB, now only low side mosfet is heating a bit. Before this pcb I did bread board tests with 2113 and saw, H side mosfet is heating while the L side was cool. Any way, I learnt that Luka is right. I think, one can solve the heating problem of L side mosfet by desinging a good layout for IR2113.
For several hours on my test I lost 3 HIP4081. But, I get experience on this chip. I have 5 smd HIP4081 now and I will desing the input sections+4081 as a module on a thin double sided pcb.
Are there any one or any idea for layout considirations on driver modules. For eg; How should be the order of components? Hıım , my prefence ; driver module then bypass caps(2200uf)+ output filter and then the mosfets. |
|
|
| luka |
Hi
Great to hear I was right :cannotbe: I did also test my ClassD with +/-47Vdc. Still works :D but can't listen for long because I dont have heatsinks for power zeners. Sound is just great, deep bass, very loud.
Oh yea, just rememberd, my friend burned IR when he didn't put enough C for high side drive. To bed hi didn't saw your post earlyer.
Well put however fits your needs/skills. You can help yourself by looking on my site under My projects form my amp, and link to Baldin's page. He has HIP |
|
|
| Whortless |
Luka,
You did not put your site's link. Would you please put it here?
Kind regards |
|
|
| luka |
Hi
It is button next to my email in every post, but just for you is here |
|
|
| Whortless |
Hi,
Luka I surfed on your website. Nice works and congragulations! I will ask a question about your car smps. I saw that U used a red toroidal core, ok do you know the part number of it? or what is the manufacturer? I have micrometals' T157-2 and I know it is for high frequency and initial permiabilty is low like magnetics' cooMu. I never tested my cores for power conversion but it is very similar to yours.
And I have a request, would you please e-mail me your class-d amp's shematic. I will only look, how did you achieve the adjustable dead time. Sorry I do not have much time to surf on web, I could not saw a shematic or a link to shematic did you put the shematic there?
About Baldin's amp ; he did full bridge but , I will do a half bridge stereo. So I prefer to use HIP4081(sperate inputs for each mosfet/dual half bridge driver) or two HIP2100(or IR2113) instead of HIP4080. My starting point is iraudamp1.
This will be a good class-d amp for my car.
Kind regards |
|
|
| luka |
Hi
Thanks.I have something like 20 of those toroinds and I dont have any info for them, just dimensions. You should use pigh power material like coolMu. I doesn't have to have low permiabilty, but its good if it has high B(flux).
On ClassD page at the bottom ther are two links, wich both are for that amp. They are links to theads on this forum. There you will find schematics for amp, and read whole thread, because there are some mistakes on schemas.
I hope I will be able to hear my amp in car in few day, can't wait:rolleyes: |
|
|
| Whortless |
Hi Luka et all.,
Ok ok. I know about saturation flux density and initial permiability. I have magnetics' ZP44413(also ZP44925) toroid for SMPS of this amp.
I know the mentioned threads but now, I understood the delay. 2113 or 2110 has shmit inputs therefore the resistors+reverse signaldiode and a cap will provide delay time. Also reverse diode ensuring fast fall time while the cap ensures lower rise time. The shmit input of 2110 corrects the inputs and finnally one can get a dead time on outputs. But, I am using an XOR CD4070 and it has propagation delay from input to output nearly >~100ns. Is this dead time not enough? And also in somewhere I saw 4069 inverters connected to achieve some dead time. 4070 or 4077 (XNOR) does nearly same thing with 4069.
Here are the data sheets :
http://www.ee.washington.edu/stores...4000/cd4070.pdf
http://www.ee.washington.edu/stores...4000/cd4069.pdf
And here is the data sheet of HIP 2100 ( my final chip for single ended module has 8ns delay maching and has internal boostrap diode)
http://www.intersil.com/data/fn/fn4022.pdf
Please have a look the data sheets and tell your ideas, may help me + any others who will DIY a class-D amp.
Regards |
|
|
| luka |
Hi
Nobody will know if 100ns will be enough, you will have to see this when you connect amp and look current that goes through fet. If there will be present some current spikes, then 100ns woun't be enough. This all depends on PCB, fets, fet drive... |
|
|
| Whortless |
| Hi to every body, Now both channels of my HIP4081 stereo class-D amp works. I am desining the smd daughter borads with different ICs. One with smd HIP4081, other one with smd HIP2100 or ISL2111(3/4A drive capability) and the last one reğresented on the below in the enclosed file with IR2113(DIP). Main board will be flexible for application of all the daughter boards. |
|
|
| Whortless |
| Please tell your suggestions about the daughter boards. They will be on a thin double sided PCB and one layer will be fully ground plane of the amp. |
|
|
| Whortless |
Please excuse me, The unused output pins of CD4070 will be removed while inserting the chip into PCB. Otherwise the outputs will damage the chip internally or may cause much power dissapation than expected.
Regards.
Ps: please excuse me for "reğresented" . The reason of this mistyping my bad keyboard and I would like to type "represented" |
|
|
| Whortless |
| If some one interested, here the finished module. |
|
|
| luka |
Hi
wow. How does it work? Man I have to do my amp in SMD too one day. |
|
|
| budget minded |
well i'm trying to create a thread looking for 25wpc & 50wpc 12v modules and have had no luck searching for them here. i don't really understand all of these technical discussions. i'm not an electronics DIYer. i'm just looking to amplify a bicycle.
despite wasting my entire library hour registering here and searching class-D to no avail, this dang forum won't let me create a thread to find the mid power modules i couldn't find online except in products lists (not purchasable)
does anyone know where i can get modules with more oomph than my sonic impact class T and with less power draw than the 100W+ modules everyone here seems to be using?
can anyone tell my why this darn forum keeps ignoring my wasted attempts to find modules other than what everyone else is discussing (at least in their post topics) module "XJ-572-QR"
means nothing to me, but 25wpc & 50w monoblock does.
i'm a layman, not a techie. i can understand what a capacitor etc. is on a schematic, but have no idea how any of it works. |
|
|
| Whortless |
Hi Luka,
I could not fully test it yet. Because I was away for weekend. But, I tested both channels with my old test board 'shown in this thread' and can say only both are working without heat on mosfets for quiscent current for an hour. I did not give audio signal because, I do not have scope yet. But I really know, from the experiences i did on dip package, the sound is good. Also a point to imply, I realised when I finished the smd board; one can do the auxilary power supplies on the empty side of daughter board. So, one can put only +/- power supplies line plus GND and audio signal and mosfet outputs to the daughter board. Any way it is working, I will desing the main board for 4 channell. I will put here the scop results when i measured. Also will desing another daughter board with HIP4080 for the fifth channel :). Hıımm, output filters; T68-2(Micrometals) 15uH for 4081 modules and T106-2 for fifth(4080) module.
Additional one thing, could someone help "budget minded"? I could not really understand the problem what he faced with.
For Luka, I will desing another daughter board with 2113(DIP package) but the input opamps and the other components will be smd and 2113 will be mounted on the same side with smd components by cutting the legs of it and not drilling the daughter pcb. I think, U can do the same for yours.
Regards |
|
|
| luka |
Hi
I will go crazy one day and put all in smd, just left IR as you said. That is because I have few of them in DIP, smd will be only to save space with resistors, caps, diods... |
|
|
| Whortless |
| quote: | Originally posted by luka
Hi
I will go crazy one day and put all in smd, just left IR as you said. That is because I have few of them in DIP, smd will be only to save space with resistors, caps, diods... |
OK Luka, What about your OSC frequency? and what are your output mosfets?
I am now desining the main board. Then I will do the second daughter board for 3rd and 4th channells. At the end I will desing the sub channel with HIP4080. I saw somewhere, HIP4080 can give 400W RMS in fullbridge topology.
Hıım, I do not know if you realised or not, the DIS pin of 4081 on daughter board. It is unused for now but at the end I'll put a protection circuit for all amplifiers. Similarly, it is possible to add protection and soft start circuit to IR2110 via the SD input.
I know, I am a bit slow, because, I am busy with chemistry :). |
|
|
| luka |
Hi
Slow?? You should look at my speed, now that is slow :D. Yes I use SD pin on IR. I have connected timer to it, wich is triggered if over current is detected.
Hip4080 will produce 400w rms easly because you can connect up to 80v to bridge, and if you use 4 ohm speaker it will put out alot more.
My freq. is 250kHz and I use not that fast Stw34nb20 fets, I mean they are fast but it would be better if I could get IRFP4227, they are somewhat better.
PS: What do you do as chemist, if that is no secret :) |
|
|
| Whortless |
It is not a secret. But, describing the work may take much time of us. I am an Inorganic Chemistry researcher . I am interested personaly with energy conversation, thermodinamic, fuel cells, polymers(inorganic+organic), catalist, etc. As everyone know, Chemistry and Physics are main science. Little developments on both may effect our life much then we thought. Energy source will be the problem of next generation and us, then we have to find more efficient energy sources. But, finding some energy source not solves the problem, for eg. only one cell of a fuel cell can produce 0.6V :). So, someone have to connect several cells as serial, up to 12V. Then, one have to convert it to 220 or any other useful voltage. SMPS' are related with this. And I know as a Chemist what could be done for power conversation. Anyway, this is only one problem and there are several scienstific problems need to be solved.
Hıım, STW34NB20 is better than my IRFP250N! My D-Amp is working with 790KHz. A bit heating :). This was only a test. On final version sw frequency wil be 350Khz.
Any way, I will measure the SW line with scop. Unfortunately could not measure yet :(. But, If i am not wrong it should be well. Because working with 330Khz there is no heat and good sound.
Regards |
|
|
| luka |
Hi
Damn yours works at 790kHz?!?.That is something. Try to find one with as little Trr - Reverse Recovery Time as possible, this will help you reach better effiecency. I had to increase dead time because of this. I don't have any heatsink on my fets for now, because they just don't "want" to heat up any more than to 40C.
Hope that you will post some pic from scope. I would like to see output ripple when there is no input signal. |
|
|
| Whortless |
Hi Luka et all.
For a long time break now I returned to my class-d amp. It is very bad to say one channel of my HIP4081 chip smoked :(. I got some scop results with the working CH but, i think the smoked ch burnt something inside the chip and the IC does not work properly. So the output sq wave is not in 50% :(. Any way my friend brought the mobile Fluke for an hour and I got these photos. More masurements will be done next weekend by changing the IC with new one.
I also tested 2113 ( 4069 for triwave+ LM311+4070+2113 and IRF640 for output) and got the scop results. The test was made on a board and all the circuit on board with single 12V supply. There are some spikes but, the circuit built on a bread board and no snubbers on output mosfet. So, it is normal I think. |
|
|
| Whortless |
I forget to say, 2113 works with 47u+100n boostrap capacity and 100u+100n on the vcc/vdd pins to ground. As shown on the pic there is dead time adjustment on the input side like 100R+SB101+1n. The circuit works cool and quite loudy for only 12V :).
Regards |
|
|
| luka |
Hi
Great, it works, well at least one does fully. What you must remember is to put alot of C for IR output drive, not that important for vcc/vdd. So 100uF for vcc/vdd is ok, but just to be save try to have 100uF normal(elko) + 22uF tantal for low an high side drive. I have that much, and what way I can't destroy my IR or fet if the amp would clip at output.
You use 4069 for what? Also I don't know if LM311 is what fast for use in D amps, well you know by now, so I guess it is.
Will you start to make boards? |
|
|
| Whortless |
Hi,
Yes I'll start but to make boards.But i gave up daughter boards. Why? Because when one uses daughter boards it is difficult to ensure short tracks between the mosfets and driver IC. But, it is usefull to make a daughter board for tests.
Did you see the output wave form of HIP4081? It is disappointing that the output has 103V pk to pk before LC filter. And 4081 only can work up to 80V :(. Although my power supply is +/-35V ,I saw this 103V on the output. My bench SMPS converts 12V to +/-35V and can handle up to 400W. But now it has a 10A projection car fuse on input. Therefore, it can not give much more than 120W or so.
Hııım my pcb component lay out enclosed. Not final, it is the last version before the final one. Final one on my home PC :(.
I will also desing a d-amp with 2113 and will make boards for it.
4069 for triwave osc ( third of inverters for triwave, see AN9405 on the link below). Plus something, I am planning to use the left 3 inverters inside the 4069 to ensure dead time instead of RDC network. What do you think?
http://www.intersil.com/data/an/an9405.pdf
Regards |
|
|
| luka |
HI
Use RDC network. Why? instead of R put trimer = pot, so you can adjust dead time for both sides. How did that 103V get there? Do you have any decoupling cap cross supply? like 470nF one? I would also use 4060 for generating triangle, I can already here you say why, well fixed freq., able to synchronize it with smps or other amp,... it is constant so you don't need to worrie that it will change during music playing....
PS: Damn board will look great, can't wait to see it... |
|
|
| Whortless |
| quote: | Originally posted by luka
HI
Use RDC network. Why? instead of R put trimer = pot, so you can adjust dead time for both sides. How did that 103V get there?
I realised that orginal supply is 70V but there is a spike down to -Vcc so it is totaly 103V measured .
Do you have any decoupling cap cross supply? like 470nF one?
There is also a 1uf MKT from + to -Vcc plus 50V-1000uf from both rails to GND .
I would also use 4060 for generating triangle, I can already here you say why, well fixed freq., able to synchronize it with smps or other amp,... it is constant so you don't need to worrie that it will change during music playing....
I agreewith you on this, but the board finished on Board maker, it requiers a bit much working on it to make all of them sync.
PS: Damn board will look great, can't wait to see it... |
Check your e-mail from me there is an interesting original shematic, from a well known audio company.
Regards |
|
|
| Whortless |
Hi to everybody,
While i am searching I found enclosed shematic. It is a subw amp and feed back line is very interesting. Half side of TL072 (5,6,7) for sq wave osc and it probably sync with amp.
All comments are welcome!
Regards |
|
|
| Whortless |
Hi to everybody,
After a while I could find some time to draw the lastest shematic. The daugther board was very helpfull because I can play with the components. I removed CD4070 and Hi pins of HIP4081A connected to +12V, so the driver can be controled from the Li pins :). But, I inverted the level shift signal with a bit wring on the board. The test circuit is more stable and fast the earlier ones and most audible from the earlier clones. The SW freq is 1MHz :). Output mosfets(4*IRFP250N) are a bit heating at this frequency, but, it is easy to get 350Khz by decrasing the integrator res. 1K to 430R.
I finished the PCB drawing of lastes version now and one of my friend will provide two original PCBs for me from an PCB company.
The amp desired for 100W+ rms power into 4ohms. I think T68-2 cores are quite at this power levels for desired power output(like Tripath's TA2022).
And finnaly I solve the unequality of output wave form. It was due to the symetry problem of power supply(+40V to -32.5V). The problem solved by adding some aux. turns to power transformer and using a seperate regulated +12V line referenced to -Vcc.
Kind Regards |
|
|
| luka |
HI
great work, I am waiting to see those boards :D |
|
|
| Whortless |
Hi, over there,
I am nearly the end of this project. I assembled the board. Unfortunately i did my pcb ot home instead of from a PCB company( waited much, one can). Any way, I would like to use micrometals T68-2 for output filter inductor. But, with this core one need to wire much more than 40sp to reach 18uH. Therefore, I found another N67 material RM6 core. Onelayer Gapless wound 8sp core gives ; 150uH and I think this is good for my output filter by givin it a bit gap on both outher and inner legs. "EPCOS' ferrite desing tool" gives 0.222mm gap, with calculation. But, I thik it is for only the inner leg. So I think, have to put 0.1 mm gap on both legs? Am I ture or are there any other suggestions? All comments are wellcome.
Regards
ps: I will put the photos of board this evening. |
|
|
| luka |
Hi
Next time use Olimex. I have just put order to them, so now I wait and hopefully everything will be OK.
Sry that I can't help you with your inductor:mad: |
|
|
| Whortless |
Ok Luka,
What core did you use for your class-D amp? Is the core seen on your website taken from an old PC power supply? If so it is probably -26 material or similar. I prefer -2 material of micrometals because it has linear permiability and also the inductance not chancing much with the current taken from the output. Gapped cores can give same permiability and inductance results therefore I decided to use RM6. And also with T68 I have to wind more then 40 sp , this would result much copper loses than expected.
regards
Ps: Wait for 2 hours! I will upload the photos |
|
|
| luka |
Hi
It is not from PC psu, but it is sam size t-106 I think it is and material should be -22, don't know really. It is yellow-red
I want pic now :D |
|
|
| Whortless |
Hi,
The file sizes are to big to show here. So, Send them to your e-mail
Regards |
|
|
| luka |
Hi
I can show them here
I have no words left... |
|
|
| Whortless |
OK, many Thanks. See the enclosed pics, one of my output filter inductor(12turns/ 19AWG ; 30uH with gap :) ) . And If I find time I'll make the measurements (if the scop will be available). But, i am sure the amp will gave a good sound with plenty of 100W*4 @ 4ohms.
Regards |
|
|
| Whortless |
| the core will be mounted between the bus caps. |
|
|
| luka |
Hi
Damn man, you should make this a bridge amp, that would be scary, drive 2 ohme load, and still have 50-70V supply
Try this someday :D:D I would very much want to know... |
|
|
| Whortless |
This is the better amp that I've built! It is working well. Sound is very good deep bass and clean high frquency. I like my amplifier :). I know that it desined only for 100Ws but, it is quite for my car.
Any way I started to class-d with enclosed UCD amp and then jump to HIP4081.
Regards |
|
|
| zerohead_ak47 |
| whortless msn mi eklermisin zerohead_ak47@hot |
|
|
|
