| cd-i |
Hello,
I want to build a power amplifier for using it at home and i need some recomendations on schematics.
I don't need something too powerful, i think something around 100 - 120 watts is enough for me. I also want the amplifier to have MOS-FET's on the output.
Browsing on the internet i found the A and T labs webpage and i think that their K9 project is ok.
http://www.a-and-t-labs.com/products.htm
If anyone know something about that amplifier because the schematic looks good at first sight. But i wanna know some ratigs about how it sounds, not only some specification and waveforms.
I should mention that i am not so experienced in audio designs (i built some amplifiers but nothing special).
Thanks in advance |
|
|
| funberry |
The K7 looks good.
What do you feed it?
I wish they had shown the bottom side of that PCB also, so we can duplicate it.
It seems like a pretty good desig, hard to beat that. There were a few MOSFET designs around the forum, look up postings by member "quasi", and they should come up.
creek amp
I'll try to dig up others.
Cum e vremea in Timisoara?
Adrian |
|
|
| tlf9999 |
| quote: | Originally posted by funberry
creek amp
|
that particular design is quite different from the K7: it is closer to the JLH in terms of the use of a phase splitter, vs. a traditional VAS there both output devices work in phase. |
|
|
| funberry |
What can I say, I'm partial to quasi-complementary symmetry, at least with MOSFETS.
The supply chain for true-audio complementary MOSFET pairs is too strangled. Scarcity, high prices, few players, high counterfeit presence, I don't want to go near that market. I stick with N-channel, all the way.
Plus there are many authors who affirm there is no real compelling sonic advantage to a MOSFET amp vs bjt designs. And I can get complementary power bjts' at prices I find acceptable.
I know, this can start a long FET vs bjt argument.
Most emerging semiconductor technologies are plagued by the same inherent challenge in audio design: The initial lack of available complementary devices. That's where IGBT's are now.
( I wish I could find a quasi-comp design using IGBT's!, these things are so rugged)
Ok here's another:
citation amp
Adrian |
|
|
| still4given |
My first amp was a "Project 101" from ESP Audio . It was very easy to build and sounds wonderful to my ears.
Blessings, Terry |
|
|
| cd-i |
hello,
Yes i like the design of the k7 amplifier too but unfortunately the layout presented on the web page, besides the fact that is incomplete, is also designed for their custom enclosure.
I have an enclosure from an old romanian amplifier and i can not use that layout. I know that a good layout is essential but i have to redesign it in order to fit my enclosure. The space is very tight in there because i intend to build 4 amplifiers in the same enclosure and power them up with the switching power supply they used in their k6 project.
Anyway this project is in a very early phase and i only have ideas of what i want to stick in that enclosure...
So if anyone can give me some tips and ideas please HIT ME !!!
p.s. hehe o surpiza placuta. La Timisoara vine toamna...
( nice suprise. The fall is coming in Timisoara...)
Daniel |
|
|
| funberry |
| quote: | | My first amp was a "Project 101" from ESP Audio |
That's a very nice amp.
The specs on it are outstanding, better than some commercial mosfet amps. It may well be the lowest distortion amp in the ESP portfolio.
Anyone wanting to build an amp owes it to themselves to consult the Elliot website, at least for some ideas.
Adrian |
|
|
| cd-i |
| yes i like the simplicity of that schematic but i have to check the availability of the power transistors in the local stores... |
|
|
| cd-i |
i also lihe the idea of building an N channel amplifier because i have some spare IRF640 mos-fets from an old project
thanks guys for helping me ;) |
|
|
| tlf9999 |
| you will not be able to use those irf devices with esp101 without modifications: you will have to replace the bias pot with a vbe/vgs multiplier. |
|
|
| cd-i |
i know that i can't use the IRF's directly on the asp101 project. That's why i like the idea of building the creek amp
I am not so good in analog audio to modify an audio amplifier and get some performances too....
That's why i prefer to build one that is designed by somebody else and is tested.
Because blowing transistors is not funny at all and after some "sparks" it's becoming frustrating.... |
|
|
| funberry |
That K6 SMPS is awesome. The price is too high for me, though.
Will you buy it, or build from schematic?
I'm using some Lambda 160W 50-0-50V 2A SMPS's to power an amp I'm working on. I'm getting these surplus for $25. Two of them in parallel, per amp. It's hard to find surplus SMPS's with useful voltages, so when I see them, I pounce.
I'm glad to see other diy-ers considering an SMPS for audio amps. I find they're such an elegant solution.
You just need to be careful with any load that may draw current peaks much higher than avarage current. A short 10A transient from a 5A X-former may dip the rail, often causing some clipping, but nothing more than that.
The same situation on an SMPS may cuse a dip in its auxiliary supply winding, which powers the primary side controller. Controllers sensing an under-voltage condition typically shut down the PWM until their supply returns to normal. The controller may wait up to a second before re-powering the primary.
Or it may go into reset, taking power from the mains resistor divider, then going through the soft-start sequence (2-3 seconds).
This can mean some gaps of silence in the middle of your music. Your guests may laugh at you, and refuse to attend your parties in the future.
I would modify any SMPS I use for audio to take it's controller supply from outside the SMPS, say from a separate, small 12V 100mA transformer/rectifier/filter cap supply.
Adrian |
|
|
| cd-i |
yes indeed the k6 power supply is very well build. I payed some attention to that schematic and i'm pretty sure that you don't have to worry about that silence in your audio program because the supply is build around SG3525 controller which is a current controller which means that the end of a conducting pulse for a diagonal of the bridge is given by the current thru the primary winding of the transformer. The conducting cycle is initiated by the oscillator. Anyway you can find details about this controller in the datasheet. And as you said the controler is powered from a separate power supply. So the only effect that i see from the schematic is a drop on the output voltage. The regulation is somehow applied to both rails in this way the unbalance in supply voltage is avoided.
And to answer to your question, I intend to build that power supply from the schematic... beacuse it's very well documentedand i have the original layout which is essential in SMPS design. If you take a look at the schematic you will see that the number of turns of the transformers are noted. And the rest is like building a kit....
but in order to be able to use that power supply i need to build an amplifier. And if i find the right amplifier for my design and measure it's power requirements i will build that power supply. |
|
|
| funberry |
Duuh.. I'm a dog.
I didn't look closely at the schematic for the K6.
It does have a dedicated controller supply.
All the inductor turns, and even the ferrite core type is specified! It's the first time I see this on a schematic. These guys must have lost their freakin' minds!
OK, I think I'm gonna build this thing.
The surplus supplies I've been fiddling with are forward converter topologies, and are prone to occasional UVLO shutdowns, but the K6 is much better designed.
There's nothing like a full bridge to make a man feel like a king!
Adrian |
|
|
| tlf9999 |
my (highly subjective) observations on the design:
a) the VAS load during the + and - cycles is not symmetric: during the positive cycles, Q4 sees a Vbe (Q7) and Vgs(Q9. for now, assumes just one pair of output devices, Q9+Q12). during the negative cycle, Q4 sees just a Vbe (Q8).
What I usually do in cases like this is to use a resistor on the emitter of Q8 to reflect the Vgs of Q12. The resistor will be of the same value of the collector resistor on Q8. as Ic ~= Ie of Q8, the voltage drop off that emitter resistor will be very close to the voltage drop off the collector resistor which is identical to Vgs. That way, the Vgs of Q12 is "reflected" on the emitter resistor and Q4 sees both Vbe (of Q8) and Vgs (of Q12).
b) I am not quite sure the purposes for those two 100pf capacitors on b and e of Q7 and Q8.
c) I still don't like the idea of having two different rails. the amp will perform identically in sonic had you had one rail at the higher voltage. it will just be less efficient (big deal).
d) I acutally prefer current mirrors over CCS, if you have to waste two resistors (I would have use neither). a current mirror load gets you equally high gains, but allows the amp to run over a much wider voltage range.
My perference is to design as simple an amp as possible. |
|
|
| quasi |
Hey tlf9999,
The cct you have linked to is the first version of many updates in a very long thread.
Please look at the final version which was developed following a lot of input by many forum contributors.
I think you'll find that the final schematic is quite different to the one you have assesed.
In any case, no attempt should be made to build the schematic you have linked to.
Cheers |
|
|
| funberry |
Quasi
thanks for the clarification.
I did instruct the reader to follow the entire threads to check for updates.
I had 1 minute to write the message and run out the door. So I delegated the thread research job to the reader.
And you did mention in your post #1 that this was an invitation for comments.
Someone just has to read.
So...
final schematic and board seem to at posts 157 and 166 of that thread.
Again, anyone reading this, read the entire thread, as there may yet be more changes.
I have a saying:
You can take a fish to water.....
but you'd better run fast, otherwise it's gonna die.
Adrian |
|
|
| wrl |
Consider building a Leach Amp.
I guess it uses BJT's but its got some of the best documentation around for how to build the darn thing. And its output is 120W per channel which seems to be what you're looking for.
-Wes |
|
|
| cd-i |
hello,
I think that Leach is a great amplifier but i'm afraid of using complementary output stages because of the transistors matching.
That's why i prefer to build something cuasicomplementary and with MOS-FET's because i have a bunch of IRF450 around and i would like to use them.
I like that NMOS schematic but the power level is too high for me,
and i wanna know if i can reduce the power level by reducing the supply voltage to something around 35-40 volts and building the power stage with only two pairs of transistors. |
|
|
| tlf9999 |
I am looking at your last version. a few comments:
a) Vce on T2/T3 appear to be too small. Ic going through T1/T5 is about 1.6ma. so voltage drop over R4/R8 is about 750mv. Voltage over R8 is about 700mv (voltage over D1/D2). As such, Vce on T2/T3 is about 650mv. I tend to run them a little bit higher (in the 1.5v range). part of the reason is high degenerate resistors R4/R8: i use 22ohm in general.
b) this is less of an issue with CCS but I usually set the current going through the Vbe resistors at about 1/10 of the current going through the whole Vbe setup. In this case, I would dial them up about 10x. Using a small capacitor here (0.1-1u) will help as well.
c) R16/R19 seem to be too big.
d) I usually make R21 identical to R22, for reasons I elaborated in my post earlier. I also used the same value for R20 and R22. IRFP140 class mosfets drop about 4ma (peak) gate current at 20khz, and I usually run my driver stage at about 15 - 20ma.
e) game resistors for the mosfets are very small. I use 110ohm to just slow down the mosfets.
Hope it helps. |
|
|
| quasi |
| quote: | Originally posted by tlf9999
I am looking at your last version. a few comments:
a) Vce on T2/T3 appear to be too small. Ic going through T1/T5 is about 1.6ma. so voltage drop over R4/R8 is about 750mv. Voltage over R8 is about 700mv (voltage over D1/D2). As such, Vce on T2/T3 is about 650mv. I tend to run them a little bit higher (in the 1.5v range). part of the reason is high degenerate resistors R4/R8: i use 22ohm in general.
b) this is less of an issue with CCS but I usually set the current going through the Vbe resistors at about 1/10 of the current going through the whole Vbe setup. In this case, I would dial them up about 10x. Using a small capacitor here (0.1-1u) will help as well.
c) R16/R19 seem to be too big.
d) I usually make R21 identical to R22, for reasons I elaborated in my post earlier. I also used the same value for R20 and R22. IRFP140 class mosfets drop about 4ma (peak) gate current at 20khz, and I usually run my driver stage at about 15 - 20ma.
e) game resistors for the mosfets are very small. I use 110ohm to just slow down the mosfets.
Hope it helps. |
a) In practice the current through R4 & R8 is about 1.3mA. I prefer to have larger value resistors here for better linearity. No reason why the amp would not work with a variety of values here.
b) I am not sure which part of the amp you are talking about, i.e. which Vbe.
c)Yes, these in fact can be removed and replace with links but T10 and T9 actually need very little drive, because there is less load on them than if they were driving a transistor output stage. The base resistors afford some protection to previous stages. But hey each to his own.
d)The value of R21 was selected during bench tweaking. I found the 100 ohm value gave the best performance in terms of symmetry, clipping and linearity.
e) I use lower value gate resistors in order to keep the drive impedance to these FETs low. The FETs used have a fairly high input capacitance. If FETs with a lower input capacitance are used then higher value resistors could be used. Again each to his own.
Thank you for your observations.
Cheers |
|
|
| cd-i |
hello guys
Again, i like this schematic but the power level is too high for me. I only need a power around 100 - 150w maximum.
My question is how can i reduce the power but keep the performace?
A solution is to reduce the supply voltage and the number of power transistors (to use only two pairs or even one pair of MOS-FETS)
Another solution is to reduce the gain of the amplifier (and perhaps the number of output transistors?)
I like the idea of reducing the gain but i don't know if that is the best solution because i have no experience in designing power amplifiers. |
|
|
| quasi |
| quote: | Originally posted by cd-i
hello guys
Again, i like this schematic but the power level is too high for me. I only need a power around 100 - 150w maximum.
My question is how can i reduce the power but keep the performace?
A solution is to reduce the supply voltage and the number of power transistors (to use only two pairs or even one pair of MOS-FETS)
Another solution is to reduce the gain of the amplifier (and perhaps the number of output transistors?)
I like the idea of reducing the gain but i don't know if that is the best solution because i have no experience in designing power amplifiers. |
Sorry cd-i
Forgot to answer your question.
If you use +/- 50 volt rails this module will deliver just over 110 watts into 8 ohms.
You will need a transformer with 38 volt windings plus a 25 amp bridge rectifier and a minimum of 10,000 uF capacitors per rail per module.
You can use the existing board layout and just use 4 FETs.
You could reduce the gain of the amplifier by changing R17. The value of this resistor divided by 1000 (R18) gives you the gain. I would not go below 22k.
Cheers |
|
|
| cd-i |
hello, quasi
I made up my mind. I will build the amplifier developed by you and presented here
http://www.diyaudio.com/forums/show...8232#post498232
Obviously I will build the schematic posted by you at the end of that thread after you worked your magic on it :D
I will also use the printed circuit board you designed for this schematic.
At first i will build the amplifier "as is" and after i see it working i will try to reduce the power.
Anyway this will take some weeks due to my school exams session that begins next week.
Thanks for your help. |
|
|
| tlf9999 |
| quote: | Originally posted by quasi
a) No reason why the amp would not work with a variety of values here. |
You want to have enough Vce to maintain linearity. at 650mv, it is dangerously close to its saturation voltage.
| quote: | | b) I am not sure which part of the amp you are talking about, i.e. which Vbe. |
Sorry. I was talking about the Vbe multiplier. |
|
|
| tlf9999 |
| quote: | Originally posted by cd-i
At first i will build the amplifier "as is" and after i see it working i will try to reduce the power. |
you can just build it, with maybe a pair of output devices. you can always parallel a resistor with the feedback resistor to dial back the gain. |
|
|
| mastertech |
funberry-
"Ok here's another:
citation amp"
man that is a Pass Classic!, these articles are awesome! |
|
|
| cd-i |
hello quasi,
yes that's right i posted the wrong link. I'm sorry but i'm a new member of this forum and i don't ktow wery well how to use all it's facilities. But don't worry i'm young and i'm lerning :D
Thanks again for the schematic. I can't wait to see it running !!!
:) |
|
|
| janneman |
| quote: | Originally posted by funberry
That K6 SMPS is awesome. The price is too high for me, though.
Will you buy it, or build from schematic?
I'm using some Lambda 160W 50-0-50V 2A SMPS's to power an amp I'm working on. I'm getting these surplus for $25. Two of them in parallel, per amp. It's hard to find surplus SMPS's with useful voltages, so when I see them, I pounce.
I'm glad to see other diy-ers considering an SMPS for audio amps. I find they're such an elegant solution.
You just need to be careful with any load that may draw current peaks much higher than avarage current. A short 10A transient from a 5A X-former may dip the rail, often causing some clipping, but nothing more than that.
The same situation on an SMPS may cuse a dip in its auxiliary supply winding, which powers the primary side controller. Controllers sensing an under-voltage condition typically shut down the PWM until their supply returns to normal. The controller may wait up to a second before re-powering the primary.
Or it may go into reset, taking power from the mains resistor divider, then going through the soft-start sequence (2-3 seconds).
This can mean some gaps of silence in the middle of your music. Your guests may laugh at you, and refuse to attend your parties in the future.
I would modify any SMPS I use for audio to take it's controller supply from outside the SMPS, say from a separate, small 12V 100mA transformer/rectifier/filter cap supply.
Adrian |
Adrian,
I just stumbled on your post. I have been using Vicor and Eta SMPS for audio power amps. What I did to prevent the sagging on load peaks is just what we used to do with linear supplies: put a hefty cap at the output. The SMPS doesn't seem to care, they will initially charge the caps with a limited current, but after that it is OK. I used 4700uF per supply voltage with no ill effects and no lock-outs. Did you ever try something like that?
Jan didden |
|
|
| funberry |
Hi Jan
Supplying the PWM controller from an external power supply has usually solved this problem for me. It's actually better if this supply can deliver some current (100-200mA), and has a large filter cap, because the controller will need to drive a MOSFET from this rail, and it may take a 1A peak to charge Cgs.
| quote: | | I used 4700uF per supply voltage with no ill effects and no lock-outs. Did you ever try something like that? |
You'r right, this definitely takes care of the problem too.
However, half of the reason I go with SMPS is to get rid of the big, heavy, expensive power transformer. The other half, is to get rid of the big, clumsy, expensive filter capacitors.
I like slim cases, 4-5cm high.
I've shoved in as many smaller, 4cm dia capacitors as I could fit, in parallel, in a previous slim design, but I still need to leave room for good air circulation.
Adrian |
|
|
| funberry |
BTW, speaking of air circulation:
When you make compact amps, you often have to provide fan cooling.
And because fans mean noise, and I like to run my fan well below rated speed.
I'll supply 7-8 V to a 12V computer fan; This shouldn't work, because the speed is electronically controlled, but it does work. At that speed, I hear no noise coming from the fan even with my ear close to it.
The lowest Voltage at which most of these fans will reliably start is 3-4 V, and I go 3 volts or so above that.
Adrian |
|
|
| cd-i |
Regarding the fan problem,
a good idea is to control the fan speed with a pwm or something similar, (if you have a microcontroler in your design) And you can spin the fan with variable speed depending of the heat sink temperature.
And with that microcontroler you can do a lot of useful things in that amplifier.
You can implement the protections circuit, a digital tone and volume control or even a remote control...
A simple PIC microcontroller is enough to do all those things |
|
|
| cd-i |
| quote: | Originally posted by funberry
BTW, speaking of air circulation:
When you make compact amps, you often have to provide fan cooling.
And because fans mean noise, and I like to run my fan well below rated speed.
I'll supply 7-8 V to a 12V computer fan; This shouldn't work, because the speed is electronically controlled, but it does work. At that speed, I hear no noise coming from the fan even with my ear close to it.
The lowest Voltage at which most of these fans will reliably start is 3-4 V, and I go 3 volts or so above that.
Adrian |
yes i agree with suppliyng the fan with a lower voltage than it's nominal value, but you may experince some problems due to dust accumulation in the fan.
Is better to use some control circuitry around that fan to monitor it's RPM and at start-up use a higher voltage for a second or two and after that you can reduce the voltage at a lower level.
You have to be careful if you use forced cooling in a tight enclosure. A failure of the fan could be a disaster for your amplifier.
Anyway I don't trust fans..... |
|
|
| tlf9999 |
| quote: | Originally posted by cd-i
Is better to use some control circuitry around that fan to monitor it's RPM and at start-up use a higher voltage for a second or two and after that you can reduce the voltage at a lower level. |
a motor does that natively: when a DC motor is stationary, it has very low resistance so a large current goes through it. When it starts to rotate, it kind of functions like a electricity generator and the current going through it goes down to the point that it is just enough to maintain the motor's speed.
I have used a thermistor to control a DC motor.
I think to be safe, any amp should have a thermal switch on the heatsink to shut down the amp in case of a disaster. |
|
|
| funberry |
Analog Devices make a neat little Temperature-to-PWM converter,
the TMP05, in SOT23 package.
I've been meaning to use them in my amp, but the dumb method works so well, I just can't get myself to use the smart method now. I wanna save the TMP05's for something else.
I have a thermal shutdown relay. The amp could have run safely without the fan, but I want to have the flexibility to place it in between books, and shelves, and other tight spaces.
Yeah, the dust, tell me about it. Once a year I have to open it up and dust it. But that's good practice for the heat sink anyways, whether or not you use a fan. A 1mm dust layer on any heat sink is destructive enough to warrant periodic removal. But I have to do that with my computer case (six fans in that one) anyways once a year, so think of it as spring cleaning.
Adrian |
|
|
| rolandong |
Hi Quasi,
I have a huge transformer intended for the super leach (65-0-65 (20A secondary), about the same supply as your high power design. I'm impressed about the simplicity and compact design of mosfets so i thought of giving it a try.
I'm willing to wait for any updates.
Just a suggestion: why not class-g output stage for higher efficiency.
rgds,
roland |
|
|
| telewatt |
| quote: | I'm impressed about the simplicity.....
Just a suggestion: why not class-g output stage for higher efficiency.
|
You won't be impressed with the simplicity any more... |
|
|
| quasi |
A class G amplifier would run off two power supplies. The normal supply and one with higher rails when more and brief power is required.
This could be achieved with this amp module. All you would need to do is track the input signal and when it got too big switch in the higher voltage supply. Sounds easy?
As Telewatt has suggested it is not quite as simple as that and the power supply and the way it feeds the amp would become quite complicated.
Cheers |
|
|
| cd-i |
hello guys,
I think that using a G class amplifier in a place where are no constraints regarding efficiency is not justified. I think that simple things are the best. Why would i complicate my life with a class G amplifier when I can use a simple one in AB class? As long as i have plenty of energy from the wall soket I don't need an efficient amplifier. I need a robust one. If my amplifier is battery powered the things change, in this case I need an efficient one. And in this case a skilled constructor can build an amplifier in class D and that's efficiency!
Anyway I think that a compromise should be made between efficiency and complexity, and this compromise is dictated by the conditions that amplifiers operates in. |
|
|
| cd-i |
I have a question related to the increased efficiency of a class G amplifier.
How is that amplifier more efficient than a usual AB class amplifier if the devices that handle the supply voltage are working in linear mode?
From my point of view this is more a way to distribute the power disipation than a way to increase efficiency.
As far as I remember this kind of amplifiers where used in car audio sistems where the supply voltage is too low to obtain high level of power, and with some switched capacitors they increase the supply voltage level when the amplifier need it. And here i'm not talking about that power amplifiers with SMPS converters to obtain 35-40 volts per rail.
Again, i'm not so skilled in audio designs, so if i'm wrong please don't hesitate to correct me. |
|
|
| quasi |
The lower the voltage across the output devices the lower the power dissipated for a certain current flow. I.e. if the average voltage across the output devices was 40 volts and the average current was 2 amps then the average power dissipated by the output stage would be 80 watts. If the average voltage was 60 volts then the average power would be 120 watts.
So it's better to run the amp with rails of 40 volts and switch in the 60 volt rail only when required. I.e to handle the very brief higher power signals.
Likewise the devices that handle the extra rails switching are on only for very brief periods and are basically turned fully on or fully off, so the power dissipation is almost zero.
Just can't get away from Ohms law.
Cheers |
|
|
| cd-i |
Now i understand where the efficiency comes from....
In my mind was a single power supply with some series regulators that feeds the power amplifier. That's why i said that is a metod to distribute the power disipation.
Now i am a little bit smarter :D
thanks quasi :) |
|
|
| funberry |
Quasi:
Since we're on class G...
Does a class G amp switch supply rails to follow a high amplitude signal cycle-by-cycle, or to follow it's envelope?
(i.e. would a 50Hz high level tone burst cause the supply to switch 100 times, or just once?)
The way I see it, unless you use an SMPS or a transformer with two secondaries ( and two complete power supplies), a class G amp simply transfers some heat from the output transistors to the power supply pass transistors. No savings in energy or in heat.
You still have to design a full fledged amp for the higher power, since in demanding situations, it may end up having to handle the higher power continuously.
Some amp manufacturers experimented with Class G some years ago, unsuccessfully --they didn't sell. All specs being equal, customers would rather buy a full-time 300W amp, than a sometimes-200-sometimes-300W amp. The public neither understood, nor trusted this power model.
Class G is still the academician's tinker toy. It's most at home on the workbench of an engineering school, demostrating some theoretical principles.
Adrian |
|
|
| cd-i |
As we were talking about class G i had an idea yesterday.
Since I intend to use a microcontroler in my amplifier to take care of some functions like protections and some diagnostics and a switching power supply (the one from the a and t labs), I could program that microcontroller to somehow control the output voltage of the power supply as function of the volume. I also intend to use a digital controller for volume and tone control but this depends on the performaces of the audio controller.
All those digital function can be easily integrated in a single microcontroller.
The advantage of using a single microcontroller that has access to volume level and depending on this can control the output voltage of the power supply by changing the division ratio in the control loop of the power supply.
Anyway the necesary supply voltage as a function of the volume level can be computed or even determined experimentally.
This may not be a very elegant idea but I think it works. |
|
|
| cd-i |
the advantage of that power supply is that it's already designed to have multiple output voltage.
Anyway changing the output voltage of that power suply is relatively easy ba changing that division ration in the control loop.
Obviously changing the output voltage thru that division ratio works in certain limits determined by the output power (upwards)
and keeping the continuos conduction mode or CCM (downwards).
Today I ordered the pcb's for the N channel amplifier that quasi recomeneded.
After I finish building the amplifier and modify the schematic for 100 - 120w, I will build that power supply and (the fun part)
play with it to obtain the maximum +/- 55 v as quasi recomended me. |
|
|
| cd-i |
sometimes I'm speaking nonsenses. I forgot to post the link to that power supply
I'm looking forward to hear some comments about my idea.... |
|
|
| funberry |
cd-i
I'm starting to build that SMPS from A&T labs now.
I have touched up the foil pattern image quality, and added holes in the center of all pads, since my drill requires this for centering. I've also separated some traces that ran too close together and could possibly short in a photographic exposure process. The pads for the IC's are also larger, to prevent foil lifting during soldering.
I've also gotten a lot of information together on the inductors and other key components.
The "T-QSE1" and "T-QSE2" part numbers are invalid. These are custom wound for A&T labs by QSE, a winding outfit in Chicago, who quoted me some outrageous prices for these parts.
Also, the value on L3, the common-mode choke is wrong; it cannot be 1.8uH, it must be 1,8mH.
If you want, we could start a thread for the construction of the K6 SMPS.
PS.
the updated foil pattern files are .PCX, the smallest lossless file type I could save in. Win Paintbrush may mistakenly think it can open these files, but it can't. You need some other graphic program to view them. I use Thumbs Plus, for viewing and printing.
Adrian |
|
|
| cd-i |
funberry,
I am interested too in building that power supply but unfortunately this month I'm stuck between my faculty exams and building the power amplifier you gave me. Anyway if all things will go acording to my plans at the end of this month I will start building that power supply too.
You were right about the value of L3, if you take a closer look at the picture of the assembled power supply you can read the value of the L3. ;)
About starting a thread, I think it's a good idea to place the thread in the power supply section of this forum to see others opinion about this power supply. |
|
|
| cd-i |
The only thing that worries me in that schematic is the current transformer. I did not see written anywhere in the documentation the number of the secondary turns.
But i think I saw the output voltage/ amp. This is enough to calculate the number of secondary turns.
Anyway I have a lot of cores around and I intend to build the inductors and the transformers. An LCR meter is enough to measure the final inductances and in case of the main transformer the leakage inductance.
Another problem for me is the pcb. I tried to make my own pcb's but the results where poor. I prefer to buy the pcb directly from the A&T or to order it to a local supplier, depending on the price.
Daniel |
|
|
| quasi |
Hey Funberry,
Sorry about the delay (I've been interstate) in commenting on the Class G thing.
Class G amps I looked at many years ago simply switched the voltage supply when the input signal exceeded a pre-set threshold and tracked the power envelope of the signal. I.e. Pre-set delays ensured the power supply did not follow the actual waveform.
I did read recently however that some class G amps use the input signal to modulate the power supply, so I am not sure whether this is done to resemble the waveform. I am not sure whether this modulation is linear or Pulse Witdh Modulation.
Cheers |
|
|
| funberry |
Here's the current transformer, D1871.
I use positive photoresist presensitized PCB's and expose them for 5-6 minutes under common fluorescent lamps (6 inches away from the lamp). For the stencils, I print on inkjet transparent sheets (for overhead projectors), using the HP' printer's heaviest ink settings ( a setting for printing t-shirt transfers puts the most ink on the page ).
I wouldn't even try double-sided boards, because they're a nightmare to aligh properly. I etch two complete single-sided boards, then stick them together, copper side out, and voila! double-sided home made goodness. ( I use bits of wire to emulate vias) The double thickness board is especially useful here, since this board is unusually large and carries lots of heavy components. I think a single-layer board would feel wobbly.
Adrian |
|
|
| cd-i |
basically my method of prototyping pcb's is the same.
But for double sided pcb's i align the printed transparent sheets whitout the pcb between them (this is easily done by moving the two sheets until the via and pad holes are mathing). After I aligned the sheets I fasten them with paper clips or staples on three sides. The fastened sheets now look like a plastic bag. You insert the photosensitive pcb in that bag and expose it.
I obtained fairly good alignment precisions in this way. My problems comes from the printer. It does not preserve the exact dimensions so I had to print 3 - 4 times the same drawing and choose the best one. Another problem was that the black colour was not "black enough". A solution to that problem was to use two overlapped sheets on the same side - and that leads me to the first problem (not preserving the exact dimensions of the drawing)...
Thanks for the datasheet of the transformer. |
|
|
| funberry |
I know what you mean.
If the printer skews the image, you can't really use it for such large size work.
Sounds like your best solution may be to take your file to a friend with a good printer.
Or if you have access to a laser printer, those can print on standard (cheap) acetate sheets. Some of them have precise enough paper transport mechanisms that you can print twice on the same sheet, and the images overlap perfectly.
If there are print shops or graphics services companies nearby, they could give you an excellent print of your artwork in notime. And they should not charge much, since doing this is really trivial for them.
BTW, do you think it would matter greatly whether one used a ferrite or a powdered iron toroid for the gate drive transformer?
Adrian |
|
|
| cd-i |
hmmm...
ferrite cores are recomended for gate drives transformers...
gate drive transformers are tricky to build.... most of the transformers I built were based on trial and error... built measured....if not worked built again....
below is a link to a very good page about gate drive transformers.....
gate drives transformers
another good link
I built many gate drive transformers based on the informations on those pages.
and related to the pcb's.... i used an HP laser printer...it was an expensive one..... maybe I will make the artworks to a print shop
Daniel |
|
|
| Gyp5y |
| Please, send me any information about how to make L1, L2, L3, T1, T2, T3, T4 from the A&T Labs K6 shematics. PLEASE! |
|
|
| funberry |
Gyp5y
As cd-i ponts out in post nr 14, all turns ratios are specified on the schematics.
If you are trying to "build" T1, or are in any way unclear about what it is, I'd strongly reccommend against taking on such a project. Any small mains transformer that puts out +/- 10 to 18 Vac at .2 A will do here.
T3 is a Coilcraft part nr, referenced in post #55.
For T4, the authors give you the nr of turns and core type.
L1 and L2 are common toroid inductors, 8 Amp wire, something between 100uH and 300uH, value is really not critical.
L3, the common mode choke, is again not too critical, since it mainly serves to protect the mains from pollution by the SMPS. It has been referenced in posts #51 and 52. You can probably substitute another common mode choke that has wire thick enough to take 10 Amps, at least for testing purposes.
If you wind your own, noone can tell you how many turns to wind, since they don't know what permeability core you will use. You may need an inductance meter to get the value right.
There is a clear risk of injury in building this, and you really have to know what you're doing.
You need to ask yourself the qustion: Can I put a scope to this thing and troubleshoot it, if it doesn't work the first time?
The manufacturer is under no obligation to provide any support information to individuals not buying their kit. They mentioned on another board that they get many novice questions, which they have to ignore, because they're not in the business of coaching or teaching electronics.
Cheers |
|
|
| cd-i |
For the transformers you'll find the informations on the schematic (number of turns, number of windings) and for the inductors you have to find suitable ferrite cores and you can compute the number of turns acording to the folowing formula:
N^2=L/AL
where
L - is your inductance
N - number of turns
AL - is a tipical value taken from the datasheet of the core. [nH/sq turn]
Anyway if you don't know what core you have you can measure and compute the AL value by winding a number of turns on the core and measure its inductance with a LCR meter. (I use to wind 10 - 20 turns. The more turns you use for that the more precision you have in computing AL) Once you have measured that inductance use the above formula to compute the AL.
Ofcourse the tickness of the wire must be chosen acording to the current that flows thru the winding. |
|
|
| funberry |
Just an FYI for anyone wanting to use original parts for T2 and T4 of K6 SMPS.
"T-QSE1" and T-QSE2 anr nor Mfr. part numbers.
"TQSE1" is part nr 486-082594, "TQSE2" is 486-060394. These are custom wound parts ordered by A&L Labs to QSE Inc.
QSE agreed to provide parts of this specification on the following terms:
(This is an email response to my quote inquiry)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Hi Adrian,
486-082594
Mosfet Gate Drive Torroidal Transformer
100 - $14.48 ea
250 - $13.14 ea
500 - $12.28 ea
1000 - $11.69 ea
Lead time 8 weeks
***3 pcs would be minimum order of $500.00***
486-060394
100 - $25.23 ea
250 - $22.94 ea
500 - $21.44 ea
1000 - $20.42 ea
Lead time - 8 weeks
***3 pcs would be minimum order of $500.00***
Can I get your company address and fax number?
Thanks
Lori
QSE INC
PO BOX 360
WATSEKA IL 60970
815-432-5281
815-432-6179 F
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Looks like I'll have to start winding...
For T2, I'll have to get some ferrite toroid cores, since all the toroids I have are powdered iron, and that won't work at these frequencies.
Adrian |
|
|
| cd-i |
The guys from A&T labs sell components from that power supply separately.
link to the price list of the k6 power supply
So if you don't wanna wind transformers you can buy them directly from them, due to the fact that those transformers are not so easy to wind and are the most critical components from the power supply. I think that buying those transformers is an option.
82$ may not be the cheapest price but still is not a fortune.
Winding those transfomers is not easy due to insulation problems at the main transformer and leakage inductance + insulation at the gate drive transformers.
I remember that I built once a gate drive tranformer for a high power welding equipment and by mistake I scratched the wire.....
huh.....new year fireworks..... those offline power supply are not kidding when they blow......
So guys.... take care..... and happy winding....:) |
|
|
| Gyp5y |
I don't understand about T4 core type& What means 28:28CT? And what core needed to make T2 and Coilcraft's current sensor?
Thaks for your replies. |
|
|
| cd-i |
...And one more thing....in the K6 power supply...
They put a snubber in the primary of T4.... R13 the resistor from the snubber is a 22w resistor....in the picture i see that they mount a wirewound resistor wich is highly inductive.... In this case I am not so sure about the role of that snubber...
As far as I know power resistors for snubbers must be non inductive. I saw some thick film resistors at digikey that are suitable for that job....
TDH35P47R0J-ND
TEH70M47R0JE-ND
The price of the TDH35P47R0J-ND is comparable to the price of the resistor they used. But it may be get hotter than the original one due to the smaller case. |
|
|
| cd-i |
| quote: | Originally posted by Gyp5y
I don't understand about T4 core type& What means 28:28CT? And what core needed to make T2 and Coilcraft's current sensor?
Thaks for your replies. |
for T4
ETD49 is the core type and 28:28ct means that the primary has 28 turns and the secondary has also 28 turns but is center tapped (14+14 turns)
For T2
a good start is to find a toroidal core with similar dimensions with the original one (not necessarily same dimensions it can be bigger...
I usually build gate drive transformers on any ferrite core that comes in my hand at the time....
In one of my early post in this thread I put two links to pages with valuable guidelines on building gate drive transformers....
and finally about the current transformer
funberry posted the datasheet of the original current transformer...
in that datasheet you have all the information you need to build one.... secondary number of turns, secondary inductance...
you can make yourself an idea of the size of the toroidal core from the outline dimensions presented in the datasheet....
for choosing the core all you need is to compute the AL value (from number of turns and inductance) with the formula I posted early and if you have the AL value and some aproximative dimensions it's easy to find a core for that transformer
:att'n: As funberry stated in post #60 you better know what you are doing there because building offline power supplies is not as easy as it looks and you need some specific equipments to debug that SMPS if it wont work from the first time (insulation transformer,circuit breaker, oscilloscope, diferential probes) and a LCR meter in order to be able to measure the inductances of the transformers you build.
If you dont have those equipments i dont see how you can build the power supply...
Anyway blowing a power supply like this one may not be fun but is definetely spectacular.... as I said .... new years fireworks |
|
|
| Gyp5y |
cd-i, funberry
I'm just want to build this amplifier. I understand (little) in amplifiers and I really don't know much in such kind of power supplies. (I'm sorry for my level of english). I wanted to build this amplifier with simple power supply (Using torroidal transformer), but I havn't enough money to buy them (I don't say about price for capacitors). This power supply is much cheaper but I have problems with finding original parts in Ukraine. So your replies is very usefull to me.
:att'n: Thank you very much for your help.
P.S. Sorry again for my english :cannotbe: (Russian and Ukrainian is much simplier to me :) ) |
|
|
| Gyp5y |
| quote: | | As funberry stated in post #60 you better know what you are doing there because building offline power supplies is not as easy as it looks and you need some specific equipments to debug that SMPS if it wont work from the first time (insulation transformer,circuit breaker, oscilloscope, diferential probes) and a LCR meter in order to be able to measure the inductances of the transformers you build. |
Don't worry. I have a such equipment (I suggest that i can get LCR meter from my friend :) ) And I'll be very carefull, 'cos like I said I really don't know much in such kind of power supplies. |
|
|
| funberry |
| quote: | I think that buying those transformers is an option.
82$ may not be the cheapest price but still is not a fortune. |
$82 is not an option for me. I cannot spend that kind of money for inductors.
After all, this is a DIY forum, for people who like to make things with their own hands.
I'm not concerned about winding T4; leakage inductance can be minimized with good , tight , symmetrical winding geometry. I may even wind one or all windings with copper tape. I've wound Tesla coils before, which require much more attention to insulation than this thing. And it doesn't have to pass CE logo requirements (most chinese transformers in computer power supplies don't either, even if they're labeled as such).
Obtaining the core for T4 may be a challenge. The ETD platform is probably one of the most expensive core+bobbin combination there is out there. The center post on the '49 is 16mm diameter, so 201 sq. mm. I think you can substiture any core with a center-leg cross sectional area similar to or greater than the ETD49.
If you kow the grade of material it's certainly better than if you don't. The K6 uses an Epcos N87 ferrite (Ferroxcube and Fair-Rite call this a 3C85) which is good from 40 to 400KHz. My rule of thumb is that if I'm going to use a core of unknown material to make an SMPS power transformer, I want a core that has served as an SMPS power transformer before.
T4 needs to be wound with the utmost symmetry. Each winding must span the circumference of the torroid, with turns evenly spaced, not bunched together on one side. Each secondary must be spread out evenly over the primary. Any assymetry here could cuase asymmetric gate drives on the MOSFETS, and a skewed transformer waveform. In worst case, the magnetization curve can be skewed sufficiently in one direction, to cause the core to slowly walk up to saturation, if the (small) dead-time provided is not enough to reset it.
If I have several toroids of unknown material I want to use for T4, I'll test them this way:
Make 2 windings of 10 turns each, insulated hookup wire. With the waveform generator, feed 100Khz square wave (through 20ohm resistor) into one winding. Load the other winding with a 1K pot ( and a 20ohm series res), and monitor waveform with a scope accross the pot. Vary the pot resistance from 1 K towards zero and write down output voltage under different loads. Repeat for all cores I have. Chose the core with the lowest drop on load in the secondary.
Adrian |
|
|
| cd-i |
hello guys,
i'm finally done with my school exams. Now it's time to build my amp.... |
|
|
| funberry |
Daniel,
Nice to hear from you.
I'm almost done building the K6 SMPS.
I have to work backwards, building the PS first, since I need it to test an amp during construction.
Adrian |
|
|
| cd-i |
hello funberry,
nice to hear that the K6 is almost done. Since I intend to build that SMPS too i'm very curios how it works. And one more thing...do you intend to use it "as is" or you intend to do some modifications in the output voltage? Because I intend to power 4 amplifiers of about 150w/ 8ohm from it. Right now i'm building the amp but i don't know it's exact power requirements (quasi's design).
quasi said that it should work at around +/- 50v but I prefer to build the amp first, see how it works and after that build the power supply. Anyway if quasi is right (and i don't see why not) i have to modify the power supply (increase a little the two 36,5k resistors R32 and R28).
Ok, enough talk for today....
I'm looking forward to hear that K6 SMPS is working well...
have a nice week-end
cheers |
|
|
| funberry |
I've tested the K6 so far at 400W output, and without feedback loop (a PWM control potentiometer temporarily spring-clipped onto the UC3525 to vary duty cycycle). Works well, and the mosfets stayed at around 65 deg C -- with no heatsink. I'm running out of good dummy loads at these powers. Next load will have to be a 1000W space heater.
The switching node waveform is not very square, owing to the poor drive capability of the controller IC. The IC can drive about 400mA peak, per output, but since both outputs are in series here, that's 400mA total, for all 4 MOSFETS. (and they all have to be driven simultaneously, 2 on, 2 off, drawing 4 simultaneous drive currents from the controller)
The high capacitance and gate charge of these devices require ample drive current (4A + per device) to produce fast switching waveforms. I'll eventually re-build this circuit with TC4422 gate drivers for each mosfet.
The IRFP450's have an obscenely high gate charge of 150nC. You can do much better than this if you go outside the IR series.
I've tried replacing them with Fairchild FQA16N50 ( 60nC gate charge), and the latter run less hot. (they switch faster, preserve a sharper waveform, and spend less time in the linear conduction region.)
Another improvement I've made was to increase the Vcc to the controller from +12V to +20V, to get more output drive. At the same time, I've decreased the turns ratio on the gate drive XF from 1.17:1 to 1.55:1. The net result is more voltage to the gate (12.9V instead of 10.25V) AND more current capacity.
I could post some pics and waveforms if others are interested in building this thing.
Adrian |
|
|
| cd-i |
hello Adrian,
I agree with you on increasing the gate - source voltage on the power mos-fets, I personally go with a voltage of around 15-17 volt on gate source. But I don't know how you intend to insert the TC4422 drivers in the circuit. They have high curent capability but they are low side drivers - you can not drive the high side mos-fets, and what about the galvanic insulation between the power mos-fets and the rest of the circuit? From my point of view it's easyer to insert a buffer between the SG3525 and gate transformer. A small buffer made from 2 transistors tied emmiter to emmiter and base to base (NPN and PNP) for each leg of the primary winding should be enough. |
|
|
| funberry |
The MOSFET drivers would go before the GD transformer. And I would probably want to use 4 transformers, one for each gate. Many drive configurations could work, both integrated and discrete; I just happen to have the 4422's from an auction bulk purchase.
The transformer would not constitute an unusual load for the 4422's, since it reflects the heavy capacitive load from the gate to the primary.
The design as it is runs very well, and I'm sure it would perform exactly as it's designers intended, given the right heat sink and fan cooling. I just think we can get this thing working with no fan and a smaller heat sink, if we can speed up the rise/fall times on the switching node. I've measured 1us rise times with the original design, brought down to about 0.5us with higher gate voltage and lower Qg switches.
I'm definitely completing this board as-is. The 4422's will require a PCB redesign, wich will come in the next build.
Adrian |
|
|
| marus |
| How do you make the main transformer T4 ? |
|
|
| jeffry_widjaja |
I suggest you build american character If you like to hear
Rap, Rock and Metal
But if you like jazz, orchestra and more specific music you must
build europian character
For me anthony e holton is more than enough is very excellent
maybe you can visit him at the website
http://www.aussieamplifiers.com/index3.htm
regards,
jeffry_widjaja@yahoo.com |
|
|
| marus |
| quote: | Originally posted by cd-i
For T2
a good start is to find a toroidal core with similar dimensions with the original one
|
Wich are the dimensions of the original toroidal core T2 ? |
|
|
| cd-i |
Hello guys
After almost two years since I started this topic I'm back.
Unfortunately I was very busy in the mean time and I did not had the chance to finish the amplifier i started to build back then. I saw that Quasi made a lot of improvements on his schematic and the thread is unbelievably long.
Anyway on last friday I finally managed to test the amplifier and like all the guys who built it I am very happy with the results. Until now I only tested it with the signal generator and the oscilloscope but the plots look very promising.
But as I stated at the beggining of this thread I need a power supply to feed the beast. My choice is still the K6 SMPS.
Anyone who finished that power supply has any opinions about it?
Today I ordered the cores to a local supplier and until I get them I could use your experience with the board.
cheers
Daniel. |
|
|
| marus |
I'm interested too in K6 SMPS.
Here some pictres with components used : http://www.yo9hjl.go.ro/smps.html
I made this in air :D ... for testing first.
But I'm afraid tu plug it in .... :hot:
Tomorrow i wil put a picture with the smps.... |
|
|
| marus |
| Daniel, this is my messenger ID : YO9HJL ;) |
|
|
| cd-i |
It seems that nobody is interested about this topic anymore..:(
I am curious to know if anybody has built the K6 power supply. I want to know if the specifications on the schematic were enough to built it or not.
Regards,
Daniel |
|
|
| viktor1986 |
What's with this topic?
Cd-i, have you built this smps power supply? |
|
|
| microsim444 |
I am working at the K6 Project. I am wondering if any one succeeded with building it?
And I cannot contact A&T labs. they dont have a valid e-mail.
Any one who suceeded into making this circuit working please tel me.
Microsim4@yahoo.com
Thanks in advance |
|
|
| microsim444 |
I was looking at the prices for the transformers, some of them is dificult to obtain some its impossible to obtain,
and AandT Labs are not responding to any e-mail messages.
I will start working into this SMPS right a way with the guids avaialble abou the GATE DRIVE transformers |
|
|
|
