View Full Version : AMP2 - experiences right out of the box
Hi Jan, hi folks,
I've received an AMP2 kit and wanted to start assembling it. But prior to do so I went through the BOM and found a few issues in there which are basically due to the fact that I'm the "poor guy" who received the "AMP2T Rev 1.0" of the PCB (with just 2 MOSFETs per channel - almost an exact copy of the original Tripath reference board design of "RB-TK2350 Rev. 2.1" far back from Sept. 2003 as label on the PCB too).
The point(s) I've to complain - taking the risk being called kinky ;) later on - are the following(s):
The provided bags with all the components do not really match with all the components to solder on this specific PCB version (to keep the church in the village - most of them do match).
An example: For the modulator gain Rfbc is splitted in two resistors (R106+R116, R109+R119 and R206+R216, R209+R219 for the other channel respectively). The Tripath' design has only (R106, R109 and R206, R209)?! That was a little bit confusing me the first time I went through the BOM (checking if all parts are delivered) and reading the Tripath docu before I read "AMP2 assembly instructions" of yours.
According to the circumstance that I've received the PCB I just hold in my hand now I decided to build up the version "RB-TK2350-2" with a supply voltage of +/-60V. With now just a 15k resistor as the Rfbc the
overall gain is just about 11.85 (poor 21.48dB) - this is far to less for the +/-60V version I'm going to realize. For the input gain stage you've provided additional resistors to select the gain, unfortunately not for the modulator stage - what a pitty. I would prefere to pay a little more for the kit (a few cent or even a Euro) instead of searching a shop (and worst, driving to a far away shop and wasting expensive fuel) where I can organize just two tiny SMD resistors ...!
I don't know how other think about it but that's what a forum is good/designed for - to discuss it.
Another component value came into my sight that deals also with the feedback loop: It's the value of C107 and C207.
In my opinion (I found nothing contradictory) the TC2001 and the TP2350 chip are identical/symmetrical when it comes to handle channel 1 and/or channel 2. But why has C107 270pF and C207 just 150pF ???
Is there any good/bad reason why they have different values? In your BOM and in all PDFs from Tripath (I double checked this) regarding the TK2350 (in all version I found today and years ago - actually the oldest is from April 2003) are both capacitors different. Was even Tripath mistaken here (possibly a copy-paste error) ???
As soon as I build up the AMP2 I'll start evaluating the "difference". To my knowledge both channel will not behave exactly the same caused by the different time constant of the low-pass filter (R106/C107 and R206/C207 respectively) fed back to the feedback inputs of the TC2001.
In the near/far future I've planned to design adequate DIY SMPS (Switched Mode Power Supplies) for Jan's AMP kits. If anyone is interested, just drop a few lines.
The advantages of SMPS are the following:
- wide input voltage range (i.e. 85...240VAC, 50/60Hz)
- regulated power supply rails for the AMP
- lack of heavy and large transformers
- lack of voluminous filter capacitors
Therefore a space, weight and volume saving method compared to the traditional method (big transformer, big filter capacitors, no regulation, ...)
I can still remember vaguely having seen a power amp in a 19 inch rack with just one unit height (44mm) capable of delivering 2x 500Wrms into 4 ohms. The power supply was build-in in the same rack, not seperately! Believe me, I was impressed. And the best - not nowadays, it was a couple of years ago and - of course - not Class-T. Just Class-D :-)) switching at stable 500kHz.
Well, that's a lot of text for now. I promise to write more "compact" next time.
Does anyone have any statements in addition to Jan's (I'm really keen on by the way - in particular to the latter two paragraphs I mentioned).
Regards
Corax
Hi Jan, hi folks,
I've to apologize for putting a question concerning C107 and C207 into the forum.
Yesterday at home I had just the reference board documentation with me.
Today - and of course after my first posting - I digged out the documentation for the TK2350 chipset itself (TK2350.pdf) and found the explanation why both capacitors should be different - how embarrassing ;)
So, never mind about this topic anymore.
Regards
Corax
Hi Corax,
If you are building AMP2, please read the forum because there was some bug on the board in version 1.0.
You are right about the "gain resistors". Let us know what values you need and Jan or someone else from the forum will just send them to you.
Designing and building SMPS for high power amps is quite painful. You can take a look at http://www.diyaudio.com for details. However, there are few companies that manufacure such PS. For example: http://www.digiamps.com . I'm not sure whether building them from scratch makes sense (in financial terms). Unless there are more people interested in buying this PS.
Marek
Corax, what did you use on D11 diode? A MURS120 ? The tripath instructions referes a B1100DICT, but i don't finde that on the components :(
Hi Markus,
as I stated in my first post I received from Jan a PCB labeled "AMP2T Rev 1.0" PCB. This one is the original design from Tripath as you can see in the corresponding manual for the reference board "RB-TK2350" and does NOT have the issue of the missing track. This PCB (with a size of 4510mil x 4175mil) is NOT identical with the PCB Jan has designed and which is (unfortunately) also labeled "Rev 1.0"! Jan's PCB has the advantage to mount a total of 8 Power-MOSFETs while the Tripath design allows only 4.
Regards
Corax
Hi jrosado,
the diode called "B1100DICT" is a high voltage Schottky barrier rectifier (as stated in the datasheet, please have a look at http://www.diodes.com/datasheets/ds30018.pdf). The reverse voltage is 100V and the forward current 1A. The MURS120 (see http://www.vishay.com/docs/88687/murs120.pdf) is rated for 200V and 1A.
In theory the little SMPS with the IRF9510 at pins 59 and 60 of the TP2350 chip will work with both diodes. The forward characteristics are almost identical.
By the way - I'm missing that particular diode too. It's listed in the BOM but not supplied by Jan :-(
But that's one of the reason why I was complaining a little bit in my first post, I just didn't mentioned everything - don't take it to personal Jan ;) you're doing (in general) a good job for the DIY community though.
Regards Corax
Hi Markus,
now I found some time in my lunch break to answer your objections about DIY SMPS.
Well, I build a few, repaired a whole bunch myself, and I'm in electronics anyway - that's my "qualified job" (or you might wanna call it "training profession") - I'm a technical engineer in electronics. So I do have enough skills to develop another SMPS. The disadvantage about the ready assembled SMPS from i.e. http://digiamps.com or http://www.alab-pro.comis the fact that they usually provide only the voltage(s) for the power-amp not for any additional auxiliary supply, neither i.e. for the +5V needed for the AMP2 nor for any other preamp stuff other DIY folks might need (+/-15V or +/-12V for some analog circuitry w/wo any linear post regulator for noise clearance due to the switched mode principles.
To add these capabilities was my intention - a single small SMPS that deliveres all necessary voltages a DIY'er might need for his special purpose. One might use the AMP as a stand alone thing while the other want the luxury version with LED-VU-meter and/or dot-matrix LCD to display volume levels and routing information about the analog multiplexers for a complete integrated amplifier.
My planned design would incorporate such things. However the specification note is not written yet, and therefore I need input from all who are interested.
My general thoughts about the specifications however are the following:
For the mains section (primary side):
- Wide input range (85V...240V at 50Hz/60Hz, or even DC, without switching or jumpering any connection(s) on the PCB).
- Always with PFC (in Germany it's compulsory for new designs to have this beginning this year, if the SMPS deliveres more or equal then 75W (if I'm not mistaken / to my last knowledge).
- Power ratings according to "worst case" conditions (maximum power output possible by design and/or selected AMP-kit from Jan).
- Very small stand-by power (+5V or +3.3V) for the (digital) push-button controlled turn-on/-off logic (i.e. via a small 1VA transformer). I can also think of a tiny µC which is powered all the time (many of them need just a few mW if in sleep mode an an interrupt driven push-button wakeup logic) to control on/off of the AMP and all pre-amp stuff.
- Manual mains switch for complete disconnect from the mains.
For the secondary side my thoughts are:
- Dual, +/GND/- selectable/adjustable supply for the bigger AMPs (AMP1, AMP2, AMP5), and single output(s), +12V...+14.4V for the smaller AMPs (AMP3, AMP6).
- Some auxiliary outputs with +5V and/or +3.3V and for analog stuff +/-6 ... +/-18V (or maybe up to +/-24V) as preregulated outputs w/wo standard 780x/790x or LM317/LM337 voltage regulators for usage on- or off-board of the SMPS.
- Overvoltage protection for all secondary voltages with crow-bars (SCR-based in connection with blowable fuses - I've seen many faulty/blown SMPS which have damaged connected devices by this issue). This would be one of the major reasons why I've choosen to design my own SMPS !!! I haven't found that feature on other SMPS yet.
- Maybe as a gadget/gimmick a voltage monitoring chip (could be inportant/fatal for dual supplies if just one of the main power rails is missing) and/or temperature measurement chip (if desired with multiple measurement points: AMP, SMPS, ...) and, to top that, a fan controller for real compact designs might be useful.
Everybody should feel free to add wishes, comments, thoughts, etc.
Any ideas are welcome.
SMPS are not that "magical" as other peoples are selling it to newbies and/or ignorants. They're basically based on simple physics and electronics. On the other hand I must admit that the theory is not always that close to the practical part. The design and test phase is, by nature, more extensive and complex than the commonly known transformer-rectifier-capacitor design but it's really not more complicated (according to my experiences).
Regards
Corax
Corax,
I have no doubts that you can design such a module. The question is as usual with simmilar projects: can you balance quality, cost and time line?
PS: the only problem that seems to me to be difficult with SMPS is so called: "power pumping". Please check the following thread on diyaudio:
http://www.diyaudio.com/forums/showthread.php?s=c3cdd7dc4b126ea7e4b3c68433b3681f&threadid=55338
Marek
Hi,
There should be an "extra" MURS120 diode instead of the B1100DICT. The ratings are very similar; ultrafast, low capacitance, and if you read the fine prints in the data sheets both pass with margin the specs for the component. The B1100 has slightly lower forward voltage but higher reverse leakage.
Yes, the two caps C107 / C207 are for setting the basic frequency of the two channels different, at least 40 kHz apart so they do not produce audible interference at idle.
If someone still has a first version of the board, with 8 MOSFET positions, contact me and I will replace them.
I still have the old board but haven't powered it up yet. Should be doing that soon. besides the missing link, is there anything else I should be aware of?
Corax,
Very nice ideas about SMPS.
I also recived my parts from Jan, and I think that I will first concentrate on PSU, before I start with soldering.
How long did you come with your SMPS? So far, I'm student [Electrical Engineering, 4th year] but if you think that I can help in any way, dont doubt to contact me. :)
Cheers
Denis
Hi atanas, hi Markus,
first of all - thanks for your interest in SMPS.
I put this into the forum to figure out the resonance from other DIYers.
As a matter of fact I made no further step about this topic. The specs I mentioned are still up and waiting for more ideas, wishes, etc.
Right now there is a huge hole in my wallet and I could not start any development no matter if I want it desperately or not. Therefore this task must wait anyway. If we both are the only interrested DIYer I have to consider to put this far below under my stack for projects to do. I am sorry about this. Primarily I am planning and building some PA-speakers (where I am going to use the AMP2 later on with). The Jensen-Imperial bass horn (just one was already satisfying) is already built. Now follows the kick-bass-mid unit with 2x 10inch plus a mid horn and a super tweeter. After I built them, the cross-over needs to be developed and finally the AMP2 is waiting for completion. Till now all components are soldered onto the PCB but that's it so far. No test run was being performed yet.
Now let me take care of the "power-pumping" phenomenon:
This phenomenon will always appear with switched-mode power amplifiers having a single-ended output. I don't wanna go to deep into detail right now (there's a lot of reading material you can find in the internet and even Tripath has dropped a few paragraphs in their datasheets) but let me highlight a few points:
1. It effects only single-ended amplifiers NOT bridge amplifiers.
2. The pumping effect will be worst the lower the frequency to reproduce.
3. The effect will descrease while increasing the bulk-capacitors of the power-supply.
Taking these 3 points into account and the knowledge that there's energy flowing back to the power-supply there are just a few things the not skilled DIYer can do about it. However some hints are those:
1. Increase the bulk-capacitors to the largest value the rectifier can handle (optionally use/replace rectifiers with higher current capabilities)
2. Use the energy flowing back in another way, i.e. in another channel. To do so you have to invert the signal to the second power-amplifier and swap the polarity of the output terminals to obtain phase relation between the two channels. However no sun without shadow: This will work perfectly only with synchronized switching amps. If both amps run on different phase or frequency it will not perfectly delete this effect.
Well, since the majority of SMPS are quiet good in sourcing current the (still) weak point is to sink current in case of the "pumping" effect. That this might interfere with the job of the PWM-controller is no secret (at least for me). Care has to be taken when designing SMPS for single-ended switch-mode amplifier - no doubt about it.
You can overcome this by selecting a different amplifier method but if your settled with a certain topology you have to live with all (dis-)advantages ;)
With kind regards
Corax
Hi Corax,
I'm ready to do the first test on the amp2 and I was thinking about the Rfbc and the gain.
I will supply my amp with 800 VA (2x42 -> 2x59.4).
With the kit I got 3 pices of 33k resistors for alt. feedback. Have you also got these?
This will give a gain of 25.53 instead of 11.85. I dont know why 33kohm? Is there any explanations? And what alternative are you using istead of 15k resistor?
Cheers
Denis
Hi folks,
@atanas:
========
First of all - I cannot remember anymore if I found these 33k resistors in my kit. But what I can tell you for sure is that I own the original Tripath Reference Board instead of Jan's 8 MOSFET circuit board.
Nevertheless, which resistors you should use depends on the two different gains you have to set up: Input stage gain & modulator (output) stage gain. Since the TC2001 (input stage) is powered with +5V only the maximum internal processing voltage (by TC2001) is +/- 2V.
According to the datasheet and the reference board design from Tripath (for the TK2350) the gains are derived by the following formulas:
Input stage gain:
Ai = -Rf / Ri = -R101 / R100 => Ai = -20k / 49k9 = 0.4 (or -8dB)
Modulator stage gain:
Am = Rfbc * (Rfba + Rfbb) / (Rfba * Rfbb) + 1 =
Am = R106 * (R105 + R107) / (R105 * R107) + 1 [for Tripath Ref. Board] =>
13k3 * ( 1k + 1k07) / ( 1k * 1k07) + 1 = +26.73 (or +28.54dB)
Am = (R106 + R116) * (R105 + R107) / (R105 * R107) + 1 [for Jan's board] =>
Am = ( 1k + 15k) * ( 1k + 1k) / ( 1k * 1k) + 1 = +33 (or +30.37dB)
The final overall gain will be:
A = -0.4 * 26,73 = 10.692 (or -8dB + 28.54dB = +20.54dB [for Tripath Ref. Board])
A = -0.4 * 33 = 13.2 (or -8dB + 30.37dB = +22.37dB [for Jan's board])
To determine the values of the resistors (R105 and R116) you have to figure out the maximum desired output voltage swing of the output, given by the rectified and filtered supply voltage (and of course the used transformer) which would be (simplified):
42Veff * SQRT(2) - (2 * Vfdiode) =
42Veff * 1.414 - (2 * 1V) = 57.4Vpeak
With a standard input voltage of 0.775Veff the overall gain needs to be:
A = 57.4Vpeak / (0.775Veff * SQRT(2)) = 57.4VpeaK / 1.1Vpeak = +52.2 (or +34.35dB)
If you use for R116 a 33k resistor your overall gain will result in:
A = Ai * Am =>
A = -0.4 * ((1k + 33k) / 1k * 2 + 1) = -27.6
Well, you need almost as twice as much gain as with R116 set to 33k. The final R116 will be calculated as follows:
R116 = ( A / (-0.4) - 1) / 2 * 1k - 1k =>
R116 = (52.2 / (-0.4) - 1) / 2 * 1k - 1k = 63.75k
Choose a value close (above) to it, i.e. 68k (E12) or 64k9 (E96).
In addition you can change R101 to 33k (was 20k) and you will gain the overall gain by a factor of 1.65 to a final gain of +45.54 - still not sufficient and with the disadvantage of a less "headroom" of the input voltage!
@jellybean46902:
I visited the site from power integrations a couple of month ago and found nothing interesting/new that would help anybody in designing a SMPS rating at 1000+ watts. I tend to design a half- or even full-bridge topology. What they offer are usually fly-back converters with a single switching transistor for SPMS up to 250W - that's in deed far away from what I dream of (at least a factor of 4 ;) ). It might be helpful when it comes to transformer selection and design, but I'll have to figure that out eventually. At least I downloaded the new version of the software.
Anyhow, thanks for your cooperation and that you give me the feeling that I'm not the only one who is involved in the "fetching" phase of this project. If you have more ideas, recommendations, etc. feel free to post them.
greetings
Corax
Hi folks,
@atanas:
========
First of all - I cannot remember anymore if I found these 33k resistors in my kit. But what I can tell you for sure is that I own the original Tripath Reference Board instead of Jan's 8 MOSFET circuit board.
Nevertheless, which resistors you should use depends on the two different gains you have to set up: Input stage gain & modulator (output) stage gain. Since the TC2001 (input stage) is powered with +5V only the maximum internal processing voltage (by TC2001) is +/- 2V.
According to the datasheet and the reference board design from Tripath (for the TK2350) the gains are derived by the following formulas:
Input stage gain:
Ai = -Rf / Ri = -R101 / R100 => Ai = -20k / 49k9 = 0.4 (or -8dB)
Modulator stage gain:
Am = Rfbc * (Rfba + Rfbb) / (Rfba * Rfbb) + 1 =
Am = R106 * (R105 + R107) / (R105 * R107) + 1 [for Tripath Ref. Board] =>
13k3 * ( 1k + 1k07) / ( 1k * 1k07) + 1 = +26.73 (or +28.54dB)
Am = (R106 + R116) * (R105 + R107) / (R105 * R107) + 1 [for Jan's board] =>
Am = ( 1k + 15k) * ( 1k + 1k) / ( 1k * 1k) + 1 = +33 (or +30.37dB)
The final overall gain will be:
A = -0.4 * 26,73 = 10.692 (or -8dB + 28.54dB = +20.54dB [for Tripath Ref. Board])
A = -0.4 * 33 = 13.2 (or -8dB + 30.37dB = +22.37dB [for Jan's board])
To determine the values of the resistors (R105 and R116) you have to figure out the maximum desired output voltage swing of the output, given by the rectified and filtered supply voltage (and of course the used transformer) which would be (simplified):
42Veff * SQRT(2) - (2 * Vfdiode) =
42Veff * 1.414 - (2 * 1V) = 57.4Vpeak
With a standard input voltage of 0.775Veff the overall gain needs to be:
A = 57.4Vpeak / (0.775Veff * SQRT(2)) = 57.4VpeaK / 1.1Vpeak = +52.2 (or +34.35dB)
If you use for R116 a 33k resistor your overall gain will result in:
A = Ai * Am =>
A = -0.4 * ((1k + 33k) / 1k * 2 + 1) = -27.6
Well, you need almost as twice as much gain as with R116 set to 33k. The final R116 will be calculated as follows:
R116 = ( A / (-0.4) - 1) / 2 * 1k - 1k =>
R116 = (52.2 / (-0.4) - 1) / 2 * 1k - 1k = 63.75k
Choose a value close (above) to it, i.e. 68k (E12) or 64k9 (E96).
In addition you can change R101 to 33k (was 20k) and you will gain the overall gain by a factor of 1.65 to a final gain of +45.54 - still not sufficient and with the disadvantage of a less "headroom" of the input voltage!
@jellybean46902:
I visited the site from power integrations a couple of month ago and found nothing interesting/new that would help anybody in designing a SMPS rating at 1000+ watts. I tend to design a half- or even full-bridge topology. What they offer are usually fly-back converters with a single switching transistor for SPMS up to 250W - that's in deed far away from what I dream of (at least a factor of 4 ;) ). It might be helpful when it comes to transformer selection and design, but I'll have to figure that out eventually. At least I downloaded the new version of the software.
Anyhow, thanks for your cooperation and that you give me the feeling that I'm not the only one who is involved in the "fetching" phase of this project. If you have more ideas, recommendations, etc. feel free to post them.
greetings
Corax
Corax,
Thank you for your post.
I also have the reference board from Tripath, rev 2.1.
The Rfbc is 15kOhm.
Another notation:
Driving the Amp on test with +/-39V the output signal has 2V without any signal, i.e. switching frequency 650kHz is making this 2V signal out.
Probably the output filter with L=11uF and C=0.22uH is not enought for this application.
Also, it can be one of typical behaviour of switching amplifiers that they make this signal out. A controversial amp (class A och AB) has much lower output noise voltage.
However according to one professor in this field, this will not effect signal quality and is nothing to worry about, but can be reduced with some modification on output filter or with adding another filter before output.
Cheers
Denis
P.S.
On normal operating, how hot is your amp (transistors)?
I still have problem with quite warm transistor (M4 and little bit M3). D.S.
Corax,
Thank you for your post.
I also have the reference board from Tripath, rev 2.1.
The Rfbc is 15kOhm.
Another notation:
Driving the Amp on test with +/-39V the output signal has 2V without any signal, i.e. switching frequency 650kHz is making this 2V signal out.
Probably the output filter with L=11uF and C=0.22uH is not enought for this application.
Also, it can be one of typical behaviour of switching amplifiers that they make this signal out. A controversial amp (class A och AB) has much lower output noise voltage.
However according to one professor in this field, this will not effect signal quality and is nothing to worry about, but can be reduced with some modification on output filter or with adding another filter before output.
Cheers
Denis
P.S.
On normal operating, how hot is your amp (transistors)?
I still have problem with quite warm transistor (M4 and little bit M3). D.S.
Powered by vBulletin® Version 4.2.0 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.