Baldin, I had terrible trouble in the beginning with both my Philips Demo PCB, and my own first prototype UCD, if I tried to connect anything less than a 3 Ohm load -- with or without any input signal!
The negative rail would 'pump' more negative, and without over volt protection (OVP), it simply took out the negative power Mosfet -- several times!!
Having read Bruno's application notes on the Hypex website, and read through some threads here, it seemed like a good idea to 'go differential' with the input -- and the feedback network of course!
Huge difference -- amps will work right down to loads of less than 1 Ohm without pumping.
If you go too low - like say a direct short across the output terminals, with no signal present - then you will still get this negative 'pumping' downward of the supply. You are of course effectively shorting out the output filter capacitor, and so I'm guessing feedback phase shift goes haywire, and the oscillator frequency and mark/space ratio go nuts on the scope!.
I should add that this happens even with my UCD180, but the OVP seems to protect it OK. On my own new prototype it protects itself most of the time!
I would suggest that making the input differential is probably the most important mod to the Philips circuit, IMHO.
The one thing I still don't really understand is why the Philips circuit includes R14 and R15 -- there have been various comments on here, but none that totally convinced me as to their real reason --- I'll bet it's got something to do with the input not being differential though!!
The negative rail would 'pump' more negative, and without over volt protection (OVP), it simply took out the negative power Mosfet -- several times!!
Having read Bruno's application notes on the Hypex website, and read through some threads here, it seemed like a good idea to 'go differential' with the input -- and the feedback network of course!
Huge difference -- amps will work right down to loads of less than 1 Ohm without pumping.
If you go too low - like say a direct short across the output terminals, with no signal present - then you will still get this negative 'pumping' downward of the supply. You are of course effectively shorting out the output filter capacitor, and so I'm guessing feedback phase shift goes haywire, and the oscillator frequency and mark/space ratio go nuts on the scope!.
I should add that this happens even with my UCD180, but the OVP seems to protect it OK. On my own new prototype it protects itself most of the time!
I would suggest that making the input differential is probably the most important mod to the Philips circuit, IMHO.
The one thing I still don't really understand is why the Philips circuit includes R14 and R15 -- there have been various comments on here, but none that totally convinced me as to their real reason --- I'll bet it's got something to do with the input not being differential though!!
Hi rogs
I see your point. Haven't tried to run it in such low impedances yet .... now know what to expect .... and will probably chicken out
Will have to consider OVP. ..... how
As to R14 and R15, they serve as a feedback path from the rails. The rail voltage will variate due to output impedance in the power supply, the two resistors will reduce this distortion. (Remember that the PSRR for a d-amp is very pour
The reason for the two being of different value in the app note is probably because they also provide a bias voltage to lower DC on the output.
I use two 1M Ohm resistors myself, and have other means of DC bias.
I see your point. Haven't tried to run it in such low impedances yet .... now know what to expect .... and will probably chicken out
Will have to consider OVP. ..... how
As to R14 and R15, they serve as a feedback path from the rails. The rail voltage will variate due to output impedance in the power supply, the two resistors will reduce this distortion. (Remember that the PSRR for a d-amp is very pour
The reason for the two being of different value in the app note is probably because they also provide a bias voltage to lower DC on the output.
I use two 1M Ohm resistors myself, and have other means of DC bias.
Hi Baldin -
I found the OVP circuit sketch from Jan Peter on the 'UCD 180 Questions' thread very useful ---if you're interested, its post #35 (on page 4) -- that will save you wading through 2000 plus posts to find it! --- (brilliant thread, but now far too loooong! )
Still not convinced about the effectiveness of those two resistors -- and although you're right about poor PSRR for 'D' amps with no feedback, I think you'll find that UcD amps have a very good PSRR.
I can't see how making them different values can accurately 'predict' the direction required by any specific DC offset - especially with an unbalanced input. I'm obviously not getting the point somewhere!!
In the end I took them out -- they certainly didn't make any difference once the input and feedback were differential.
I found the OVP circuit sketch from Jan Peter on the 'UCD 180 Questions' thread very useful ---if you're interested, its post #35 (on page 4) -- that will save you wading through 2000 plus posts to find it! --- (brilliant thread, but now far too loooong!
)Still not convinced about the effectiveness of those two resistors -- and although you're right about poor PSRR for 'D' amps with no feedback, I think you'll find that UcD amps have a very good PSRR.
I can't see how making them different values can accurately 'predict' the direction required by any specific DC offset - especially with an unbalanced input. I'm obviously not getting the point somewhere!!
In the end I took them out -- they certainly didn't make any difference once the input and feedback were differential.
How to drive TO-247 mosfet, like IRF250N?
With TO-220 mosfet, my CCT oscilates at 300khz. Changing the mosfet to IRF250N, the oscilation only 100khz, this mosfet is "hard" to drive.
How can I get the same 300khz with IRF250N?
Is it possible to make Q12-D11 and Q11-D12 to totem-pole / changing D11-D12 to NPN transistor? Will it help? Is R17-R18-R8-R21 need to be re-adjusted?
With TO-220 mosfet, my CCT oscilates at 300khz. Changing the mosfet to IRF250N, the oscilation only 100khz, this mosfet is "hard" to drive.
How can I get the same 300khz with IRF250N?
Is it possible to make Q12-D11 and Q11-D12 to totem-pole / changing D11-D12 to NPN transistor? Will it help? Is R17-R18-R8-R21 need to be re-adjusted?
stepheno said:
hi
stepheno
if u like ,u can ask me on my new post
MCD-II
MCD-II:http://www.diyaudio.com/forums/showthread.php?s=&threadid=96501
Regards
fumac
Why is that my UCD clone has a "bird" hidden inside a cct, making intermittent chirping sound? It's annoying, I wanted clean reproduction without any bird sound within the music.
Is there anything wrong with these traces?
Residual at idle 1V36 (p-p).
Why is the residual has 2 blips on left and right of sinusoidal?
Is there anything wrong with these traces?
Residual at idle 1V36 (p-p).
Why is the residual has 2 blips on left and right of sinusoidal?
Attachments
Where the chirping sound comes from? I couldn't see anything in the scope.
My UCD will blow mosfets if played loud. Why is that? With IRF540 it just 1 second to blow. With IRF640 5 seconds to blow. With FDP3672 5seconds. Before the mosfets blowing, the sound is distorted much first, then the mosfets blown away. When the sound is distorted, the current flowing into the cct rises very much.
What is the mosfet failure mechanism when the amp is playing loud? My amp doesn't have any protection at all. Gate resistors are 22ohm. Rail voltage = +/-45V, speaker =4ohm.
My UCD will blow mosfets if played loud. Why is that? With IRF540 it just 1 second to blow. With IRF640 5 seconds to blow. With FDP3672 5seconds. Before the mosfets blowing, the sound is distorted much first, then the mosfets blown away. When the sound is distorted, the current flowing into the cct rises very much.
What is the mosfet failure mechanism when the amp is playing loud? My amp doesn't have any protection at all. Gate resistors are 22ohm. Rail voltage = +/-45V, speaker =4ohm.
I would say that you have a combination of high peak currents and poor layout. The blips after the filter come from EMI going through the output filter, the overshoot of your square wave comes from poor layout of the inner current loop of the switchers => inductance.
You can lessen the effects of this by switching softer resulting in lesser current peaks drawn through the FETs, but to what extent depends on your design and layout altogether.
You say your waveforms are fuzzy is this with no signal driving the UcD? The UcD will shift frequency when driven, so you can only observe the cleanness of a signal with no input drive.
You can lessen the effects of this by switching softer resulting in lesser current peaks drawn through the FETs, but to what extent depends on your design and layout altogether.
You say your waveforms are fuzzy is this with no signal driving the UcD? The UcD will shift frequency when driven, so you can only observe the cleanness of a signal with no input drive.
Hi, Zilog,
All the pictures I attached is condition with no load. I guess this can conclude to 1 thing, if we want to make UCD, SMT component is a must, where we can have short tracks everywhere. Using ordinary components will lead to long tracks everywhere.
Before the mosfets blown, the sound becomes distorted first, and the current draw is rising very much. Why is this? Is there a shootthrough between top and bottom mosfets?
All the pictures I attached is condition with no load. I guess this can conclude to 1 thing, if we want to make UCD, SMT component is a must, where we can have short tracks everywhere. Using ordinary components will lead to long tracks everywhere.
Before the mosfets blown, the sound becomes distorted first, and the current draw is rising very much. Why is this? Is there a shootthrough between top and bottom mosfets?
Ther could be many things wrong to make the mosfets burn.
SMD will for sure be the better choise, but I don't think this is what's wrong.
The ringing on the mosfets (what you call blips*) can be dampned with snbbers. Try 10 ohm in series with 1 nF over each mosfet, and see on the scope whether this helps.
The bluring on the scope can as Ziglo says be variations in the switching freq, or just som higher harmonics of the the carrier freq. Remember that the residual is not a single sinus but a series of sinus tones making up the square wave.
The bird sond could be some parts of the circuit (think output stage/filter) emmiting EMI that couples into the amp again.
Try to post schematics, the layout and some pictures of the amp .... otherwise it will be dificult to help you further
SMD will for sure be the better choise, but I don't think this is what's wrong.
The ringing on the mosfets (what you call blips*) can be dampned with snbbers. Try 10 ohm in series with 1 nF over each mosfet, and see on the scope whether this helps.
The bluring on the scope can as Ziglo says be variations in the switching freq, or just som higher harmonics of the the carrier freq. Remember that the residual is not a single sinus but a series of sinus tones making up the square wave.
The bird sond could be some parts of the circuit (think output stage/filter) emmiting EMI that couples into the amp again.
Try to post schematics, the layout and some pictures of the amp .... otherwise it will be dificult to help you further
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.