An amplifier such as a Class D design is super critical on the physical layout and all the component interconnections so that interference and instability do not occur.
They are very unforgiving and unfortunately the end result is as you discovered 🙁 but you have to then ask questions and learn... and if that basic interest is there then you will.
Last post from me tonight 🙂
The slew rate... I think what he is saying is that compared to a valve stage that can only change its output voltage relatively slowly, the Class D stages swing their output stages from rail to rail in microseconds and less.
So the rate of change of voltage vs time (the slew rate) is massive compared to a valve. That in itself can cause big big problems because bits of wire don't quite behave like you think when you try and couple those kind of signals into them. And that causes problems.
So there is nothing to be embarrassed about, and as you read up on it all you will realise a different skill set and rules come into play for some aspects of high speed high current design in just the same way that very high frequency RF design plays by different rules 🙂
First, thank you for taking the time to actually answer my question.
When I read the an-1135.pdf file that talks about pcb design, I didn't understand it to say 'you can ONLY use pcb' but more to be 'thoughts on how to design where things will be to maximize performance' Not 'if you don't use pcb and do everything exactly as we state, your ic will let out the magic smoke'
Some notes I took from that document stated that the input NFB is one of the most sensitive parts, especially the section from the left of the resistor into the NFB pin. For that reason I put the resistor itself closest to the input I could and ran the jumper wire for it UNDER the FR4 material which is very thick, hoping this would remove the potential for noise. Also the part they referenced as the most noisy, is near the NFB but on the top side instead, again by my thought process, to avoid interference.
These are possibly mistakes coming from the Class AB Tube amplifier world where placement of components is also important especially in high gain situations, but my understanding of lead length and component placement is from that world.
I also made sure the mosfets were down and to the right to try and keep them away from any input level signals due to knowing they are higher current devices that would also create noise.
If I'm to understand the only way to build one is via a PCB due to the need for extremely short leads and extremely separated components etc, is there a PCB someone has that's available that I can find/buy etc?
I like building things, and had hoped I could do this myself, but I definitely have spent a lot of time reading/watching videos on pcb design and know for a fact I have no clue how to do it. Nor do I really want to spend the 50000 hours it will take to truly understand and master it. This was part of why I tried prototyping it on FR4 with eyelets.
also per the note "Class D- this is very short wire, in your layout very long" is a statement that gives me more questions, but doesn't answer any... why must class d wires be shorter? is it capacitance coupling? How short? how thin of traces? Why?
Again, I'm hoping to learn why, or I'll keep making the same mistakes. I didn't expect anyone to handhold me through this, i spent possibly close to 6 months prepping for this prototype, failed, and then happily came here hoping to learn. Failures are how I learn. I felt instead insulted, and berated for being so lame and stupid... I hope that makes sense?
I am the admin and moderator of a forum for tube amp builders, and am decently aware of many electrical problems, characteristics and methods, but rarely have I been treated like this for asking for help before. Again, thanks Mooly for treating me with respect.
Pompeiisneaks, sorry you got mostly such lame replies after I suggested you post here. I haven’t been particularly active here for some years and it looks like many of the crowd I knew have moved on, and it doesn’t seem hard to see why...
irs2092 is very fussy about short pcb tracks and decoupling.
decouple close to 2092 and close to mosfets.
its fussy about which mosfets you use.
large mosfet gate capacitance will trigger overload protection.
for heavy mosfets use tc4420 gate drivers
deadtime needs to be right so you get no shoot through with mosfets.
if the mosfets go they usually take 2092 with them so change both.
use mosfets used by IR datasheet to start with.
can you show how vcc supply is generated ?
decouple close to 2092 and close to mosfets.
its fussy about which mosfets you use.
large mosfet gate capacitance will trigger overload protection.
for heavy mosfets use tc4420 gate drivers
deadtime needs to be right so you get no shoot through with mosfets.
if the mosfets go they usually take 2092 with them so change both.
use mosfets used by IR datasheet to start with.
can you show how vcc supply is generated ?
VCC was separately provided via my bench power supply with - connected to B- and 12VDC connected to the pin 12 line. Rigol PSU bench supply.
Thanks for the info, I guess swapping the mosfets wasn't a smart idea. The problem with the ones they suggested are SMD types and I was, as seen, trying to do a through hole style build. I'd love to find someone with pcb's using those SMD style mosfets and for the SMD style IC itself, then I could build it myself, get that satisfaction of it being done by me, etc. 🙂
At any rate, it looks like so far as I'm hearing my way won't work.
Thanks
Thanks for the info, I guess swapping the mosfets wasn't a smart idea. The problem with the ones they suggested are SMD types and I was, as seen, trying to do a through hole style build. I'd love to find someone with pcb's using those SMD style mosfets and for the SMD style IC itself, then I could build it myself, get that satisfaction of it being done by me, etc. 🙂
At any rate, it looks like so far as I'm hearing my way won't work.
Thanks
back to basics a valve amp is low frequencies compared to class d.
while a class d carrier is around 150khz the edges of the wave forms go into megahertz.
this means any pcb track small inductance/capacitance has a big effect on the carrier causing delays and ringing. to minimise this pcb tracks should be very short and fat as possible.
while a class d carrier is around 150khz the edges of the wave forms go into megahertz.
this means any pcb track small inductance/capacitance has a big effect on the carrier causing delays and ringing. to minimise this pcb tracks should be very short and fat as possible.
Last edited:
No quote button for some reason...
Indeed. Some time ago at work I was helping with R&D on a PFC boost reg and we were switching off about 15 amps in 30nS. IIRC we were getting a 5 volt spike in the TO-247 mosfet source lead from pcb surface to mosfet plastic body, a distance of about 4mm. Of course, this spike effectively appears in series with the gate drive signal, causing unexpected results. So, long skinny tracks are a definite liability.
Makes more sense now, so a lot like SMPS with the high frequency stuff? that's starting to make sense. Again, I've never dealt with this kind of high frequency ringing etc, due to everything I did being in the audio spectrum, but I get how Class D works and uses high frequency as well.
Hi pompeiisneaks
basically answering your general problem:
1) Class D amps are VERY fussy about layout
2) not a good starter for people coming into Solid State from a different background, personally I´d build a couple "regular" projects first, good old Class AB amps, classic "Iron" supplies, etc.
Zillions of projects to choose from, I´d start with a popular and well documented one, so you have ample reading material which will often answer questions you didn´t even ask 😱
3) you mentioned "not building a kit" .... why not?
It can be very good practice. (still build a couple Classic projects first)
4) absolute worst case, at least use a commercial board (you supply and assemble everything else) which has been used by others and is a proven design.
5) whichever path you choose, keep asking here 🙂
basically answering your general problem:
1) Class D amps are VERY fussy about layout
2) not a good starter for people coming into Solid State from a different background, personally I´d build a couple "regular" projects first, good old Class AB amps, classic "Iron" supplies, etc.
Zillions of projects to choose from, I´d start with a popular and well documented one, so you have ample reading material which will often answer questions you didn´t even ask 😱
3) you mentioned "not building a kit" .... why not?
It can be very good practice. (still build a couple Classic projects first)
4) absolute worst case, at least use a commercial board (you supply and assemble everything else) which has been used by others and is a proven design.
5) whichever path you choose, keep asking here 🙂
JMFahey,
Thanks, I've built a lot of tube amp projects from both kits and from scratch sourcing components myself. I understand what makes good components. I chose WIHA caps for this because I know they're high quality. I know that rubycon, nichicon, etc are great electolytics. I wanted to try and make a simple class D solid state amp due to reports that they're quite a bit nicer sounding than solid state Class AB etc.
I'm okay with a kit for this, my pride wanted to be able to do it myself 🙂
If you know of a great class D kit, let me know!
~Phil
Thanks, I've built a lot of tube amp projects from both kits and from scratch sourcing components myself. I understand what makes good components. I chose WIHA caps for this because I know they're high quality. I know that rubycon, nichicon, etc are great electolytics. I wanted to try and make a simple class D solid state amp due to reports that they're quite a bit nicer sounding than solid state Class AB etc.
I'm okay with a kit for this, my pride wanted to be able to do it myself 🙂
If you know of a great class D kit, let me know!
~Phil
As i told you in first replies, you cannot do such amplifiers by just connecting wires. Class d generally works above 200khz. You not only required a good layout pcb but also an oscilloscope. Everyone is a noob untill he tries and fails.
Specially for Irs2092 you should follow the application notes. And if you want to fast forward everything then just build a pcb using iraudamp7 gerber files. If you do everything as mentioned in there then it will work right from the first time.
Hope it helps you
Specially for Irs2092 you should follow the application notes. And if you want to fast forward everything then just build a pcb using iraudamp7 gerber files. If you do everything as mentioned in there then it will work right from the first time.
Hope it helps you
I've never built a Class D amp, kit or otherwise so I'm afraid I don't know of or have any recommendations on that front... perhaps some of the other Class D enthusiasts might though.
When talking about very high frequencies 'long' can be just a few 10's of millimetres (sometimes very much less).
A totally different area but think of RF design and a rooftop aerial and its coaxial downlead. It is just a copper inner conductor and copper or other metallic outer conductor... or is it?
If you stick a pin through the cable shorting inner to outer what might happen to the signal at the receiver? You might be surprised to see it actually increase on some channels, disappear on others and leave some unchanged.
It is because that metal pin, which is a dead short at DC and low frequencies behaves totally differently at very high frequency. It possesses not just stray capacitance but also self inductance that comes very much into play at the frequencies involved interacting and modifying the signals flowing through it. It creates unwanted tuned and resonant circuits, alters the phase of voltage and currents... which in a high power switching Class D stage could be disastrous.
And so the same applies to the construction of the Class D amp. Even short PCB traces will behave in unexpected ways. Sometimes the shape of the PCB trace such as a using a curve to do a 90 degree turn rather than a straight right angle can make a difference in how currents flow.
Thanks for the kind words 🙂
Keep reading, keep asking and also hopefully someone might have some ideas on something you could build, kit or otherwise.
I've built a number of the LJM L15D IRS2092 mono and stereo kits available on eBay (e.g. here) or Aliexpress (e.g. here). There may be other kits too, but I only have experience of these ones. They're 150W into 8 ohms, or if you want more power, the L20D gives 200-250W but I recall might be a bit sensitive to speaker impedance.
I'm currently using seven L15D's in my home theatre power amp, powered by 2 x 500W +/-55V SMPS, and the sound quality is excellent to my somewhat ageing ears.
LJM's designs are quite close to the IRS2092 reference design, although they omit some of the protection circuitry. The PCB's are excellent quality, and the IRS2092 is pre-soldered to the board, but the rest of the components are through-hole that you solder yourself.
And if you manage to blow up one of these boards (which I have done a number of times accidentally), replacing the IRS2092 and mosfets isn't too much of a challenge - I keep a stock of them in the parts bin for just such an occasion!
If you haven't already, it's worth reading the IRAUDAMP7 reference design and if you're building a multi-channel amp, check out the sections on bus pumping (use an inverting buffer on alternate channels) and switching frequency. On my boards, I've replaced the resistor that controls switching frequency with a multi-turn pot for fine adjustment.
Hope this helps.
the 2092 is only available in smd now.
a pain if it blows.
i have a cheap hot air work station which removes them fairly easily.
a pain if it blows.
i have a cheap hot air work station which removes them fairly easily.
I've actually got a hot air station too. I've made some pcb's that are 16 pin mount with through hole for them. I just realized via this mess they'll be useless for this purpose, now to find another good use for them 😀
I'll check out those kits and probably do one. I want to use it for a power amplifier for a tube based preamp in a guitar amp... in theory.
So single channel, up to maybe 200W max is fine. Probably even better if it's way lower like 50W and can be played at bedroom levels or sent off to an audio interface from the preamp for recording etc.
Thanks,
~Phil
I'll check out those kits and probably do one. I want to use it for a power amplifier for a tube based preamp in a guitar amp... in theory.
So single channel, up to maybe 200W max is fine. Probably even better if it's way lower like 50W and can be played at bedroom levels or sent off to an audio interface from the preamp for recording etc.
Thanks,
~Phil
All the ones I've built have been SMD, but not hard to remove and replace with just a fine tipped temp-controlled iron if you're careful.
I can't remember seeing the LJM kits in mono, only stereo (two boards). And for 50W or less, if you're not heart-set on the IRS2092 and/or class-D, you could also maybe look at some of the chip-amp kits on ebay or Ali.
Also, the IRS2092 reference design has a pretty low input impedance, around 3.3K if i recall correctly. Not sure if this is an issue - if so an op-amp buffer between preamp and main amp might be needed.
Last edited:
If you're after a lowish power classD kit then in a day or so I'll be opening sales for my design using a TDA8932 : Transformer input TDA8932 mono amp