| Arius |
Hello,
I'm happy to say that I've completed one channel of my OnSemi ThermalTrak amplifier. It is essentially a modified Leach clone (many thanks to Jens).
The amplifier is completely modular with onboard PSU and protection. Just plug in the transformer, signal input and speaker outputs. It measures 13.3" X 2.95" - the goal was to make it a compact module to enable me to build a 5-channel amplifier.
Picture:
(sorry, both the camera and my camera skills suck)
 |
|
|
| Arius |
Here is a picture of the bare PCB...
 |
|
|
| Arius |
I don't have access to a distortion analyzer so with a HP 33120A Function Generator and an Agilent Infiniium scope, I took some measurements.
Some scope waveforms....
Yellow = Input signal
Green = Output signal across dummy load
Purple = PSU +ve rail |
|
|
| Arius |
| Another scope waveform: |
|
|
| Arius |
| Schematics attached.... |
|
|
| Arius |
I must add that I haven't been able to fully characterize electrically the Vbias circuit based on the ThermalTrak diodes. It works but so far, I can't tell if it is doing a better job than a plain old VBE amplifier transistor.
I have access to a thermal chamber for testing electronic products but I'm not sure if heating up the entire module and measuring Vbias proves anything. It's the junction temp vs Vbias that really interests me.
Other than that, the module's was completed yesterday and it sounds very good. Can't say too much as it is hard to compare a mono module vs my stereo hifi setup. Work is going slow on the 2nd module... |
|
|
| BlackMTX |
| Nice PCB layout,compact... |
|
|
| clem_o |
Hi Arius,
Why not just measure idling current before and after giving the amplifier a 'good exercise'? If the idling current dominates over the rest of the amplifier, that should be a pretty good test. Most amplifiers drift all over the place, specially from startup when everythng's cold...
Cheers!! |
|
|
| acenovelty |
Hello Arius,
That's a really fine lookin' layout. Maybe you would be willing to tell more about the circuit and construction details. Like the purpose of a relay on the board and other small details.
How does the bias adjustment compare to a Leach board and so on.
Inquiring minds want to know.
Regards |
|
|
| Arius |
Hello Ace,
Thanks for your inquiry. Um, I did post the schematics here so all the details are there.
The relay is for DC fault, turn on/off, overtemp protection. Bias is adjusted just like a regular Leach or any other regular Emitter-Follower class b amp.
There's really nothing special about this amp other than the ThermalTrak's diodes reacting faster to temp changes. Like Clem suggested, I think I'll use my test setup and 300W dummy load to run the module through various levels of input signal and see how the Vbias tracks temperature changes. It won't be accurate as I'll be measuring Vin vs Vbias instead of Tjunction.
BTW, my toroidal transformer is Antek Inc's 700VA type. It has pretty good load regulation. Rail voltage dropped from 53V to 50V at full load. With 50V rails, I get Vout = 44V before clipping so that's 120W into 8R. Heatsink is a Conrad MF25-75.
The PCB is double sided with ground planes. Noise/hum is inaudible even with the input floating. The purists probably won't like to see the PSU circuit onboard, but my goal is for a compact, modular amp to enable 5 channels - minus the wiring nightmare. To remove one module from the chassis, there are only 3 connectors to unplug (4 if one chooses to bring the status led to the front panel) and 2 screws.
If anyone wants any detail, just ask away. Just keep in mind I'm not an amplifier expert. Just know a bit here and there. I'll start work on the AC soft-start/soft poweron board next. |
|
|
| acenovelty |
Noticed that the internal diodes for Q11,23 are not attached to the string. What might be the logic/reasoning for this?
Regards |
|
|
| Stocker |
That is a very good looking getup you have there. I am impressed and I hope to see more commentary and especially other members trying it for themselves. I have always been a fan of the Leach amplifier, I wonder how the sound of this compares...?
Anyway, I like the modular concept and you get 0 points deducted for taking a long time making the 2nd channel, we all know how that goes. ;) |
|
|
| lineup |
| quote: | Originally posted by acenovelty
Noticed that the internal diodes for Q11,23 are not attached to the string. What might be the logic/reasoning for this?
Regards |
It is probably because then the bias circuit would have gotten too high voltage = too much bias.
Each diode will add like 0.6-0.7 Volt.
The idle current bias in output is set by the voltage between Q5 BASE and Q25 BASE.
Depending how you design this bias circuit and type of output stage
you may only need to use 1-2 diodes in some cases.
(if you only have one output pair, you will only have 2 diodes)
Those two free diodes can be used for other thing.
For example a LED indicating a certain temperature,
either that amplifier has reached working temp at switch on
or a red warning LED for overheat. |
|
|
| acenovelty |
Are these imbedded diodes different from the 1N4004's on the Leach LowTim amp? That amp and the Superamp use only 4 diodes. Prof. Leach mentions that 4 only made it convenient for them to come out on the same side of the heat sink.
Here Arius has used 3 diodes per side(6 total) and 4 transistors per side(8 total). Just trying to make sense of the logic or math involved.
The amp looks very impressive and well executed.
Regards |
|
|
| Arius |
Lineup is correct about me not using all 8 of the ThermalTrak diodes. Reasons:
1) 6 diodes is good enough to set Vbias accurately. Simulations confirmed this. It did not overcompensate (I think).
2) 6 diodes is what OnSemi had used in their appnote. So I could bypass the VBE transistor and just used the 6 diodes in a string to set Vbias - no adjustment.
3) I did not want to be forced to use 4 pairs of output devices all the time. 3 pairs is sufficient to deliver 300W into 2ohms. The 4th pair was added at my friend's (band player) insistence.
I regarded the embedded diodes as equivalent in function to the 1N4004's in the Leach amp.
The last two embedded diodes are unused. The 4th pair may be dropped (ThermalTrak's are cheap at OnSemi but pricey at Digikey).
There is onboard thermal cutoff and the status led will flash if that occurs. I thought of using one of the last diodes for temp sensing but I didn't want to deal with an opamp and its power supply hassle.
I'm still sniffing around for a spectrum analyzer. I'll also post more about sound quality when the 2nd module is finished so I can listen to a stereo setup.
One last thing:
Digikey = $7.43
Future = $6.00
OnSemi = $1.85
Being a "frugalphile", I only embarked on this because OnSemi was willing to send me samples although I told them outright it was for a personal project. Don't see how they're gonna sell many of them via Digikey. Hmph! |
|
|
| clem_o |
| quote: | Originally posted by Arius
I'm still sniffing around for a spectrum analyzer. I'll also |
Send me a board and I'll do the distortion analysis for you! :-) :-) :-)
Clem |
|
|
| jackinnj |
| quote: | Originally posted by Arius
I'm still sniffing around for a spectrum analyzer. I'll also post more about sound quality when the 2nd module is finished so I can listen to a stereo setup. Hmph! |
Looks just great !!!
Too bad EBay T&E has dried up -- watch for the Tektronix DA4084 or AA501 analyzer and the SG505 signal generator on EBay -- they often come up as a duo and will work to around 0.0007% THD. You can also do a good job with the TrueAudio soundcard software. |
|
|
| clem_o |
Dont quite remember what model I have, but its a Tek DSP-based distortion analyzer. Coupled to a Tek DSO plus some software for FFT, does useable job!
Cheers
Clem |
|
|
| lineup |
| quote: | Originally posted by Arius
2) 6 diodes is what OnSemi had used in their appnote. So I could bypass the VBE transistor and just used the 6 diodes in a string to set Vbias - no adjustment.
........
One last thing:
Digikey = $7.43
Future = $6.00
OnSemi = $1.85
Being a "frugalphile", I only embarked on this because OnSemi was willing to send me samples although I told them outright it was for a personal project. Don't see how they're gonna sell many of them via Digikey. Hmph! |
I found this OnSemi PDF Application:
Single Channel Audio Amplifier with 3 pairs of ThermalTrak
MJL3281A / MJL1302A driven by
MJE15032 / MJE15033 and predrives
MPSA06 / MPSA56
and a differential NPN input stage.
Power supply is separate +-55 Volt for Input/VAS and +-45 Volt for Power output stage.
This will naturally improve performance.
http://www.onsemi.com/pub/Collateral/BRD8043-D.PDF |
|
|
| peranders |
| Arius, good work. :nod: Looks nice :up: |
|
|
| Maniac |
Just wondering, can I buy a few boards off you? Or any chance of getting the PCB file to have the PCB made locally? This board seems to be rather nice and compact, and might just be perfect to drive some speaker that my current gainclone just can't handle :P
Thanks |
|
|
| Arius |
Howdy,
I haven't updated this thread in a while. Nothing new to report really. Finished the 2nd amplifier module. Listening to the stereo amp lying naked in my listening room and it sounds great. I'm really slow with the casework. Just started work on the soft-start.
Currently, I do not have any spare PCB's available. I have 2 extras but am holding on to them for a friend. So if you are interested in building your own, I can supply you with the exact same gerber files that I used for production. The design works as published, no rework, part changes, etc.
The gerber ZIP file is 280kb. So I can't attach it here. Email me arius(at)surewest.net for a copy. I authorize reproduction for non-profit personal use only (it is a hobby, after all).
By the way, each board has a total of 8 output devices. That's some money there. |
|
|
| peranders |
| Anything to report? |
|
|
| EWorkshop1708 |
It's nice that the last 2 diodes are available for other circuits. Sweet PCB design.
I imagine a fan control, or a temp protection shutoff for the amp would be a great use.
I've been tempted to try a pair of thermaltraks as drivers for a large number of regular output devices to see how well that would work, since they would warm up as the heatsink warmed up. |
|
|
|
