Power amp under development

Jump to Latest
#1,941
danieljw said:


Quasi

what sort of voltage limitation does the NMOS 350/500 have, is it based solely on the r6 and r17 combo and the two 2sc1845 until Vce limit of the MJE340/350s is reached ?
Click to expand...

The answer to this is quite complex, but I'll give it a go. Remember though that I am not an engineer (Q is a salesperson).

The more obvious answers have to do with SOAR, availability and price of suitably rated components (transistors capacitors etc.) plus the issue of heat in the varoius stages. There is also safety.

Then there is another interesting issue that has received some thought but probably not enough. I note that AndrewT may have given some comment about this some time ago and I'll go over this at the end of this text.

So lets cover the more obvious ones;

1. SOAR - finding suitable devices and then keeping them within a safe load demand and then keeping them cool. This is easily solved by more or stronger devices and bigger heatsinks. Unfortunately for an amp with a FET output stage this often means a higher gate input capacitance and this places a greater demand on the driver transistors as they try to charge this capacitive load. So the drivers have to be upgraded.

2. As the voltage increases, finding low noise and high gain transistors for the first stage becomes more difficult. Plus you have to allow some headroom to allow for power rail fluctuations. An alternative is to use a cascode stage, but this approach is not my preferred method.

3. The heat generated within the stages starts to become an issue. When you have almost 100v across a component even a few milliamps of current can be an issue. With 80 volt rails the second and third stages dissipate around 5 watts and almost all of this is in the 4 MJE transistors. This is a lot to get rid of.

4. With +/- 80v rails consideration must be given to safety. In terms of mortality 160v DC could be enough to challenge it.

Finally and this is the less obvious bit. The amp is a minimalist design that places great demands of the components in it. Most designs that deliver this type of power will generally have an extra gain stage, or darlington or other "compound transistor stage" to provide sufficient gain for good results in terms of linearity, distortion and noise. This amp does not use these techniques (one of my design principles is less is better). This provides a limit to the overal gain and subsequent power that can be delivered faithfully.

Cheers
Q
 
#1,942
Thanks Q,

I appreciate your detailed response.

U probably guessed i am one of these power mad people
tying to push the envelope a little too much.

actually I am just trying to decide wether to build the actrk600 or the nmos500 to mod for some high power output.

i like both designs but i am having a tough time choosing which will be easier to mod.

by mod i mean run +/-95Vd.c. and a "truckload" of output stage mosfets 🙂

-Dan
 
#1,945
Hi Quasi,
Came across this thread ten days ago (took me that long to read the whole thing right from the first post!). Its a fantastic job all you guys are doing here.
I'm planning to upgrade from my mini crescendo power amp and the nmos 350 looks like a very exciting prospect.
My plan is to use 2 pairs of IRFP450 with a supply rail of +/- 65V DC into 8 ohms speakers. I'm also working on modifiying the nmos200 TO247 PCB for the nmos 350. I'll post the design once its ready.

Hari
 
#1,946
Re: Reducing T6, T7, T9, T10 Power Dissipation

thierry said:
hi quasi !
thanks for your job and your answers.

i've realized my own pcb with your schemetic becaus my heatsinj and the free place are special.

it's working but i got some probllem with quiescent current and voltage bias between t8 CE.

the first time i 've used some igbt for output devices,it s working

but the thermal stability is awful, quiescent current go up and down dramatically so i've changed igbt for mosfet (fqa 44n30 fairchild )

the fist time i started the amp with +-45 v temporary supply
i ve changed r13 to 1200 ohms to get a correct voltage for Vgs (about 5 volts to run 100 ma per mosfet )

but after a few moment,i ve checked the small heatsink and i ve found it very hot with this low votage supply (+-60 v will be worst )
so i decided to change current across r15 (now it s 100 oms )
and i ve changed r20,r22 to 510 ohm to reduce power dissipation in t6,t7,t9,t10.
ok temp seems to be correct (35 ° with 20° ambient )

but now i cannot get mote than 3 volts across r20 and r22

do you know if the new current in t7 is too low or i have to change some resistors values ?
Click to expand...


quasi said:
Hi thierry,

Does your amp work. I mean have you heard clear sound through a speaker.

If it does work then may I advise the following.

Changing R13 to 1200 ohms will not help the heat situation much and will alter the characteristics of this section. I suggest you put the 470 ohm resistor back in. The small heatsink normally runs very warm, almost too hot to touch, but the MJE340/350's are designed to take it. Take the time to look at the datasheet and you will see that these transistors can still dissipate a lot of heat at high temepratures.

But if you are worried about the heat you can increase R12 & R15 to 68 ohms, these must be the same. This will decrease the second stage current by 3.5 mA and will reduce the heat produced by T6 & T7 by 0.4 watts.

The only way to reduce the heat on T9 & T10 is to increase R20 & R22. I suggest using 330 ohm resistors to reduce the current through these by 6mA. The power dissipated by these transistors will then drop by about 0.6 watt.

So in total that's a saving of around 1 watt and it may be enough to make you more comfortable.

Once this is done reset the bias using VR2. I only use 30mA per FET pair, in my opinion there is not enough benefit in running 100mA, but it's a personal choice. Remember though that 3 FET pairs with 100mA bias current each on 60 volt rails equals 36 watts continous dissipation, so even large heatsinks will get quite warm.

Anyway, thanks for choosing this design. If you have the time please send me a photo of your work.

Cheers & Good Luck

Quasi
Click to expand...


quasi,

The above are from:

Post #336 19th January 2006 06:42:18 by thierry

Post #337 19th January 2006 07:24:25 by quasi

The question I have is in your reply of Post #337 was the suggested values for R12/15 and R20/22 in the context of the base design that uses +-75 VDC rails or in the context of +-60VDC rails that thierry was using?


Regards,

John L. Males
Willowdale, Onario
Canada
05 July 2007 07:43
Official Quasi Thread Researcher
 
#1,947
jethari said:
Hi Quasi,
Came across this thread ten days ago (took me that long to read the whole thing right from the first post!). Its a fantastic job all you guys are doing here.
I'm planning to upgrade from my mini crescendo power amp and the nmos 350 looks like a very exciting prospect.
My plan is to use 2 pairs of IRFP450 with a supply rail of +/- 65V DC into 8 ohms speakers. I'm also working on modifiying the nmos200 TO247 PCB for the nmos 350. I'll post the design once its ready.

Hari
Click to expand...

Hi Hari,

Are you aware of the 2 pair variant PCB I created in the "classical" NMOS350/500 layout? Are you going to use some sort of DC Protection circuit? I know you read the whole thread which is great. Your motivation to enhance the NMOS200 TO-247 version for 2 pairs? I am just curious as I am sure you learned reading the thread. 🙂


Regards,

John L. Males
Willowdale, Ontario
Canada
05 July 2007 08:26
Official Quasi Thread Researcher
 
#1,948
Re: Re: Reducing T6, T7, T9, T10 Power Dissipation

keypunch said:






quasi,

The above are from:

Post #336 19th January 2006 06:42:18 by thierry

Post #337 19th January 2006 07:24:25 by quasi

The question I have is in your reply of Post #337 was the suggested values for R12/15 and R20/22 in the context of the base design that uses +-75 VDC rails or in the context of +-60VDC rails that thierry was using?
Click to expand...


The values of these resistors set the current through the relevant stages. Beacuse heat = V x I reducing either voltage or current will reduce the heat. Because we're stuck with the voltage we need to reduce the current. Now whilst the transistor selected in these stages can take the heat some constructors will find it troublesome. So for rails above 50 volts and if the heat worries you make the changes suggested.

Cheers
Q
 
#1,949
keypunch said:


Hi Hari,

Are you aware of the 2 pair variant PCB I created in the "classical" NMOS350/500 layout? Are you going to use some sort of DC Protection circuit? I know you read the whole thread which is great. Your motivation to enhance the NMOS200 TO-247 version for 2 pairs? I am just curious as I am sure you learned reading the thread. 🙂


Regards,

John L. Males
Willowdale, Ontario
Canada
05 July 2007 08:26
Official Quasi Thread Researcher
Click to expand...
Hi John
No, I did not notice that layout. I'll be glad if you could provide the link or the actual .pdf layout. I'm only using the NMOS200 PCB layout as I prefer to use the angled bracket mounted to the heatsink. This is pretty much the same way the mini crescendo is designed. That way the heat sink remains at the back of the case.
I do plan to use a DC protection circuit. I might go for the quasi circuit or stick with the one I'm using right now which is an old Elektor design. By the way, does the quasi circuit also provide switch on delay?
Hari
 
#1,950
jethari said:

Hi John
No, I did not notice that layout. I'll be glad if you could provide the link or the actual .pdf layout. I'm only using the NMOS200 PCB layout as I prefer to use the angled bracket mounted to the heatsink. This is pretty much the same way the mini crescendo is designed. That way the heat sink remains at the back of the case.
I do plan to use a DC protection circuit. I might go for the quasi circuit or stick with the one I'm using right now which is an old Elektor design. By the way, does the quasi circuit also provide switch on delay?
Hari
Click to expand...

Hi Hari,

The PCB versions I have been creating variants on are the "classic" PCBs which are for the NMOS350/500 amplifiers. These PCBs designs are different to the NMOS200. The former typically has the PCB parallel to the heatsink and this generally results with the heatsinks on the side of the case. This is not the orientation you are looking for.

I have not made any attempts effecting modifications for the NMOS200 for a few basic reasons - I like the larger PCB rail filter capacitors and larger NFB to gound filter capacitor, has the DC Protect on the same board as the amp module, and it lends itself to some other "creative" modifications I have explored. The downside of the NMOS350/500 PCB is T9 and T10 is not mounted on the main heatsink, whereas with the NMOS200 T9 and T10 are mounted on the main heatsink. I suspect it really would not take that much effort to get T9 and T10 for the NMOS350/500 to be on the main heatsink. That said the NMOS350/500 works just fine heatsinked as shown on the PCB and many picks will show as you read the thread.

The advantages of the NMOS200 is smaller physical size, the ability to use an "L" mounting bracket, ability to mount a number of amp modules side by side. The NMOS200 does not have any DC Protect on the amp module board like the "classic" NMOS design has.

The DC protect circuit has been published as a seperate PCB, but there was a rather recent thread review and rework proposed. I have not had the time to translate the proposed DC protect changes to the schematic and PCB. Many are using the original PCB/Schematic for the DC protect which is the same as the on on the "calssic" PCB.

If you are still interested in the 2 pair version of the "classic" NMOS, AKA NMOS350/500 let me know. If you really need to use a right angle bracket you may be bias (pun intended) to the NMOS200 TO-247 version modified to your 2 pair need.


Regards,

John L. Males
Willowdale, Ontario
Canada
06 July 2007 00:52
Official Quasi Thread Researcher
 
#1,951
keypunch said:


Hi Hari,



If you are still interested in the 2 pair version of the "classic" NMOS, AKA NMOS350/500 let me know. If you really need to use a right angle bracket you may be bias (pun intended) to the NMOS200 TO-247 version modified to your 2 pair need.


Hi John
My going for the NMOS200 layout is precisely for that reason - so that I can use the right angled bracket and mount the boards against the rear panel of the case. This is primarily because I go in for integrated amplifiers (including a phono preamp section). As I mentioned, I'm only using the PCB layout of the 200 - the schematic would still be that of the 350. In fact, the two schematics are almost the same but for some component values.
What about the switch on delay in Quasi's DC protect circuit?
Thanks
Hari
Click to expand...
 
#1,952
jethari said:
keypunch said:


Hi Hari,

What about the switch on delay in Quasi's DC protect circuit?
Thanks
Hari
Click to expand...


Hari,

I am not certain if the Quasi DC protect circuit has a switch on delay, but as you are aware the Quasi DC protection circuit was the subject of some review a few months ago while you read this thread. The proposed alternate has a switch on delay. That said, but quasi will need to confirm as I am not an EE or near quasi's knowledge, I have a feeling the Quasi DC protection circuit also has a switch on delay. I have not been able to confirm this in this thread, but I seem to recall some questions the odd time regading changing the timing of the Quasi DC protect circuit. I have not been able to find this information in the thread as my time has been limited on what I can spend on thread research I am sad to say.


Regards,

John L. Males
Willowdale, Ontario
Canada
07 July 2007 11:41
Official Quasi Thread Researcher
Click to expand...
 
#1,953
John
I too remember vaguely reading about the switch on delay. I certainly dont feel like reading the whole thread again :smash: We'll just have to wait for the master 😀 to respond!
BTW I've just finished my PCB layout, please do take a look at it for any flaws.
Thanks
Hari
 
#1,955
Hi Hari,

jethari said:
John
I too remember vaguely reading about the switch on delay. I certainly dont feel like reading the whole thread again :smash: We'll just have to wait for the master 😀 to respond!
BTW I've just finished my PCB layout, please do take a look at it for any flaws.
Thanks
Hari
Click to expand...

I am sure the "master" will know the answer in his sleep.

jethari said:
John
The pdf file I created is 144kb and the forum is not accepting this size as an attachment. I'll work something out in the meantime.
Hari
Click to expand...

Yes, I have had that problem with the forum as well. The limit is 100k. That said depending on what tool you use to create a PDF it may be possible the tool or the option of the tool was not set to compress the PDF. What are you using to create the PDF? Now I am in my "element" (IT/Software).


Regards,

John L. Males
Willowdale, Ontario
Canada
07 July 2007 15:34
Official Quasi Thread Researcher
 
#1,956
keypunch said:



Hari,

I am not certain if the Quasi DC protect circuit has a switch on delay, but as you are aware the Quasi DC protection circuit was the subject of some review a few months ago while you read this thread. The proposed alternate has a switch on delay. That said, but quasi will need to confirm as I am not an EE or near quasi's knowledge, I have a feeling the Quasi DC protection circuit also has a switch on delay. I have not been able to confirm this in this thread, but I seem to recall some questions the odd time regading changing the timing of the Quasi DC protect circuit. I have not been able to find this information in the thread as my time has been limited on what I can spend on thread research I am sad to say.


Regards,

John L. Males
Willowdale, Ontario
Canada
07 July 2007 11:41
Official Quasi Thread Researcher
Click to expand...


The original DC detect circuit does have a switch on delay offered mainly by the 330K resistor plus the 470uF capacitor. The timing can be adjusted by varying either of these component values within a reasonable range. The schematic is on my web site.

Cheers
Q
 
#1,958
Re: Performance report

Samuel Jayaraj said:
This is how I configured the final setup of my Quasi Amp:

1) X-former : 0-45, 0-45 @750VA (+-63 volts DC)
2) 2 bridge rectifiers
3) Caps : 7500uF x 2, 0R1 10watts, 7500uF x 3, 0.1uF x 2 per rail
4) Rails cut between Front End and Driver/Output stage - inserted IN4007 diode, 10R 2W resistor, 470uF + 100uF + 0.1uF
5) Changed R4/R8 to 330R
6) Changed R6 to 10K
7) Changed R12/R15 to 680R
8) IRFP450 x 3 per bank
9) KSE340/350 instead of MJE types
10) C4 is Silver Mica
11) Zobel Network is 10R 5W + 0.1uF 250V
12) Iq is 100mA
13) C1 = 11uF currently. I will have to pop in a 1uF plastic type.

Now the amp is really upto the challenge. Rhythm and pace are among the very best. Very good resolution of low level signals. Sound stage is very controlled and not aggressive or harsh. Low Bass and Mid Bass are very, very good.

Does not image as forward as my AKSA clone, Stochinno Amp and Gain Clone (LM3886TF), in that order; however, the mids are not withdrawn or shy. I would have liked the overall image to be a little more forward, personally speaking. But the musicality and timing more than make up for this.

T6/T7 do run a bit hot - have to increase the heatsinking for these.

Any comments or opinions?



E and R both represent Ohms.
Click to expand...

Samuel,

I have a some quick questions for you as it was my intent to split the rails between the output devices and driver stage. As I mentioned before I also have the KSE devices rather than the MJE and the toroids I have are 44-0-44. Because of of these similarities I wanted to ask you these questions:

1) Regarding your point (4) I assume the "470uF + 100uF + 0.1uF" are in parallel. The questions are why do you choose a high value 470uF and why parallel two electrolytics of different vlaues?

2) Regarding point (4) Is the 10R 2W resistor in series with the supply rail?

3) Regarding point (4) is the IN4007 used to from the supply line to ground to protect the front end/driver stage from the rail reversing polarity?

4) Any specific reason why you would not use a small fuse to also protect the front end/driver stage from the supply?

5) Might it be possible for you to make a quick schematic of just the components of item (4) so I am clear about the IN4007 and 10R 2W resistor. I am certain of the "470uF + 100uF + 0.1uF" part of the configuration, but please inclue them in the quick schematic. Some

6) Any reason(s) you felt setting Iq to 100ma per point (12) rather than the suggested 30ma?

7) Re points (5), (6), and (7) do you think the values you changed for these resistors was related to you using the KSE devices rather than the MJE?

8) Do you have any thoughts why your experience with the stock values and rails of +-35V or +-42V did not have the issues you had using +-56V and +-63V based on the changes you made?

Samuel, thanks for making the postings you did about your experience and what you did to resolve the issues you experienced. In addition to splitting the rail at the driver and output stage I intended to do before youe postings I am leaning towards using the Leach type of NFB in the version of the quasi I am building. I am hoping at worst doing so will not negate the the many happy builders with the quasi design. If I feel it does, it will be easy enough for me to revert back to the quasi NFB design.


Regards,

John L. Males
Willowdale, Ontario
Canada
07 July 2007 21:28
Official Quasi Thread Researcher
 
#1,959
jethari said:
The master has spoken!!!😀
Quasi, thanks for the reply. In that case I'll go for the quasi DC protect circuit.
John, I'm creating the pdf file directly from protel dxp. I'm still trying to get a more compressed file.
Hari
Click to expand...

Hari,

Will Protel output a Postscript file? If it does and using ideally bzip2 with option 9, or second choice zip with the -9 option does the resulting compressed Postscript less than 100K? If the compressed file is less than the forum 100K limit post the compressed Postscript file and I can see with the tools I have if I can create a PDF under the 100K forum limit.

As FYI, I am sure you are a Windows user and so will say what the heck is bzip2? bzip2 is a compression program like zip, rar, arj, et al. bzip2 will compress a file better than zip will. bzip2 is open source and as such you can download it and use it with no fee to pay. It is a command line program, but it is really easy to use: bzip2 -9k [name of the file to compress]. The resulting file will have the same name/extension as the file being compressed with ".bz2" added as the file name suffix meaning the file extension will now be ".bz2".


Regards,

John L. Males
Willowdale, Ontario
Canada
07 July 2007 21:44
Official Quasi Thread Researcher
 
You must log in or register to reply here.
Top Bottom