Discrete audio amplifier project PT12 circuit board layout idea

[@LunarHermit]

I LOVE the idea of mounting everything onto an aluminium plate! Drilling the holes in the heatsink is one of the worst things about assembling amp kits in my opinion. I also love the idea of everything being a compact module of sorts. Really excellent idea! You could also have the holes on the plates tapped for nicer machine thread screws rather than self tappers. I'd happily pay extra for both options! Also gotta love the Snickers at the end~

[@JohnAudioTech]

Thanks for your comments. I never use self tap screws on heat sinks.

[@rrangana11]

I like your arrangement of that heat sink for the driver, power and the biasing transistors, this arrangement will track well and also makes very compact compared to an "L" arrangement. Great work John. I strongly believe this amp will become very popular, so please do consider to publish this in Audiokarma, Hifivison etc of DIY places.

[@MarkTillotson]

Looking good - quite taken by the flat layout, could be very sleek in the right enclosure. One thing at the end, "difference", not "differentiate"...

[@DAVIDGREGORYKERR]

I idea revolved around the Wide-band Class A amp of which there were three which drove the color guns in a color CRT.

[@savvassidiropoulos5952]

Nice simple circuit. If I may suggest, research the VAS stage a bit. The simple transistor will be a weak point with plenty distortion in the high frequencies. Check for EF-VAS topology, it will be a lot better. The layout looks like you have plenty of space there to add a transistor and resistor.

[@JohnAudioTech]

I will certainly consider this change if I'm not meeting my numbers. The VAS stage is the first place to start.

[@andymouse]

I have found several data sheets with different pinouts for the 2N5401C device, but I see you have ' EBC' so I will stick with that ! also as of today Digikey don't have them in stock and many configurations of the device are obsolete, but I have yet to try Mouser, who supplied them to you please ?....cheers.

[@micomrkaic]

John, using scope and sound card for testing your amp is not either-or. You could use a decent inexpensive sound card and a piece of software like ARTA to get a good measure of your amp. ARTA also allows you to test intermodulation distortion and can give you precise numerical values for THD etc. Your scope does not give you any of that. If you want to look at higher frequencies with your scope, you can still do that.

[@twotone3070]

Would there be some advantage in using a double sided board but only using the reverse as a ground plane for shielding?

[@nitescuvalentin6872]

Neat design! Looking forward to buy the whole nine yards package (most likely)

[@JohnAudioTech]

Thanks! I hope I can do this for a reasonable price.

[@christophschuermann6512]

Hi John, nice video. Please have a look on the analog discovery analyzer as an affordable measurment system. It can measure up to 10 MHz and has a 14bit resolution.

[@JohnAudioTech]

Thanks for the tip. I'll check it out.

[@iblesbosuok]

Mounting power amplifier directly on heatsink is good idea. Beside get shielding, you will get mechanical toughness. I love your hood-cat, I love all kinds of cat. Cheers from Indonesia

[@sandro2304]

Hi John, not sure if someone commented this on a future video, but you have a mistake in the input network. You do not need the 1K resistor on the base of Q2. It is only giving you an offset. The base current of Q3 flows through the 22K resistor attached to its base, but not the 1k due to the capacitor. This creates a voltage drop on the 22k resistor which needs to be balanced, Vdrop = 22k*IB3. The base current of transistor Q2 flows through the 22k and 1k resistors attached to its base. Since the resistors are in series, you get a voltage drop of 23K*IB2 . You will see because of this a DC offset of ~1K*IB. I can't remember your numbers (would need to back track too much), so let me make some assumptions here. If: - IC_Q2 = ~IC_Q3 = 4mA - Q2, Q3 Beta (hfe) = 100 You will see an offset of -IC_Q2/Beta*1K = -40mV. The fix, remove the resistor... easy. Hope this helps, Sandro P.S. It is good practice to also label resistors.

[@JohnAudioTech]

Hi Sandro, as it stands now, the amp has -33mv offset. When I bypass the 1k resistor on the base of Q2, the offset changes to -40mv. The LT spice model shows +20mv with the resistor and +15mv without. The collector currents in Q2 and Q3 are 1.3ma each. The 1k resistor is a base stopper to help reduce RF ingress. Many amp designers include it at the cost of some offset. As far as why my circuit offset is so much, I'm not sure but Q3 did suffer a base to emitter reverse bias for several minutes due to a soldering error and that may have affected its gain.

[@sandro2304]

@@JohnAudioTech Hi John, the 1k resistor on the base of Q2 is unnecessary for RF stopping. You already have a 1k forming a low pass filter with the 220pF, so the extra 1k will just give you noise (4nV/rtHz) and offset as I pointed out. Now, given your IC=1.3mA and still assuming a beta = 100, then the Delta offset will be -1.3mA/100*1k = -13mV. Since you see 5mV in LTSpice and 7mV in the actual circuit, then Beta must be higher than my assumption (~200 rather than 100). BUT, the direction is correct, i.e. offset goes in the negative direction, showing my analysis is correct. There are other sources of offset in this circuit, some easier to fix others harder, that contribute to the -33mV offset you are seeing. By design, you have 20mV ofsset as stated by LTspice. We know 5mV is due to the unnecessary 1k resistor. The rest is due to the difference in base current of Q9 vs. the sum of the Q4 and Q5 base currents. To eliminate this one, you need would need to run Q9 at 2.6mA (IC5 + IC6) assuming they are the same transistor model. On top of the systematic, you have -53mV of offset due to matching of devices. This is due to mismatch of Q2 vs. Q3 (VBE and BETA), Q4 vs. Q5 (VBE only), and mismatch in the associated resistors. You can mitigate this using match pairs and resistor networks. Yes, these cost some money, but if this is a hobby project, then extra bucks are worth it. If you plan to sell this amp, then maybe not worth it. Hope it helps. Best, Sandro Another piece of feedback, label the transistor models (e.g. 2NXXX).

[@JohnAudioTech]

@@sandro2304 Thanks for the reply. You convinced me to drop the 1k base resistor in the final design. sound-au.com/amp_design.htm section 1.2 explains the reasoning for having but it looks like the bad effects outweigh the one benefit. I would be content if the offset is under 25mv in the final design.

[@sandro2304]

@@JohnAudioTech Hi John, thanks for pointing me to that website. Some good info there. What Elliot describes is the rectification effect of AM and he correctly points out that this is mitigated in op-amps by putting the 1k resistor. What the 1k does is create a LPF with the input capacitance of the op-amp. You have probably heard that this technque works best with JFET input op-amps... well, that is because signal JFETs have bigger capacitance than BJTs , no magic =). Also, try to place to 220pF near the base connection of Q2. It is will minimize any antenna effects from the trace.

[@wallybaldemor8523]

Why use current mirrors instead of just resistors? Thanks

[@JohnAudioTech]

Higher open loop gain.

[@Marco.SalaMessom]

Hi there From Brazil, O can't wait for the amplifier,since the First vídeo.

[@cocosloan3748]

Woow Finally nice decent amp we can all make!!

[@mrmax1467]

Hi John, I’m new to Ohio, I heard you mention having aluminum pieces machined. Is there a machine shop near Dayton / Yellow Springs area that will work with small runs? Also, do you happen to have any pcb’s for sell? I’m looking to build an IC chip amp. I love watching your videos, thanks.

[@amitanaudiophile]

Hi very informative videos you uploaded sir. Sir my interest is is there any hi fi IC available which has great output result as least IM distortion and autobias of output stage.

[@PelDaddy]

Dixon Ticonderoga No 2 pencil FTW!!!!

[@KuntalGhosh]

can u make a review of the tda7850 boards? , I can't find any good video with proper testes ..... most are just repairing old radios or testing it with some stupid song and speaker without putting out any actual values that I am interested in like distortion and RMS Power ... without clipping .. or any sort of noise ... none of the video shows it in details... ,

[@bixy9347]

I really hope it works John, it’s a great project. After watching today I started thinking what other designers subscribed thought about your design. Maybe you could reach out to them, have a Skype conversation live stream, I’m sure they would be happy to talk with you. Just my thoughts, love your channel 👍.

[@MrBrymstond]

As long as that plate heatsink can dissipate the heat fast enough at 4ohm high volume, it's a great idea

[@jakep8484]

well you would mount the plate to a heatsink with some compound

[@MrBrymstond]

@@jakep8484 Of course, but I meant a plate of metal vs a heatsink with fins and a possible fan. I'm all for the the plate if it will keep it cool.

[@CliveChamberlain946]

IMO the PCB's should be double sided with fully masked ground plane on the top side with plenty of via stitching everywhere down to grounds on the underside. This will keep it noise-free as a mouse - even Snickers will be happy! Seriously, even if you don't take the time to run this flood-fill through a free program like KiCad, the price difference for a PCB is inconsequential these days compared to the component cost of the whole project. Sorry for the long babbling.. ;)

[@natecontarino1748]

It would be nice if you could get a little Snickers face logo thingy in the copper or silkscreen.

[@tc-bladeofgrass6719]

Have you considered a kick starter or something similar to take a few pre orders for the kit and use that money to buy your audio analyser ? It wouldn't take any effort to get the required 700 dollars. I'd pre order for 2 kits right now 😁

[@boots7859]

John, did you ever post a BOM? I, like many others probably am looking for a simply, decently powered amp for a garage shop environment. I'm thinking this might be overkill for your average MP3 player/streamer compared to something like a TDA3116D2. PS, your Thumbs Up vs keyboard EE/H8ers is pretty good.

[@JohnAudioTech]

No BOM yet as the amp is still under development. I'll know if the complete circuit even works at all soon.

[@natecontarino1748]

Will this play Crysis?

[@JohnAudioTech]

No but I might be able to get Donkey Kong to run on it.

[@イエンスヨハンセン]

You’re a clever bloke - very clever - why don’t you make your own differentiating circuit ana make your own distortion analyser?

[@isoguy.]

Awsome, great project. Would love to see bluetooth capability added to project rather than having to add another board.

[@kindanyume]

Never add that to the power amp /facepalm... that would be an optional input method for a preamp and even then BT really is NOT a good method for hifi use for many many reasons.

[@vicentecortez8636]

Love all your videos!! What would be the price for this amplifier?

[@JohnAudioTech]

That is an unknown at this time. I'd like it to be around $50 US for the whole kit w/ aluminum plate.

[@walterbaldemor2128]

5 mica insulators for outputs and drivers

[@anindyamitra5091]

Did you see earlier, the protection circuitry that STK41xx use? You can follow this datasheet pdf1.alldatasheet.com/datasheet-pdf/view/41598/SANYO/STK4141.html They have also provided the internal circuitry of the chip.

[@ejlockpix]

John I could help you measure the distortion at low levels. I have access to high end equipment. Let me know if you want to chat about it.

[@johnyang799]

How low can you go?

[@ejlockpix]

If you refer to distortion I'm not sure but I have scopes that can go up to 8GHz with 20GS/S or more, 10-16 bit ADC. Spec an that goes from 30Hz to 6.5GHz. We have higher frequency spec ans that are like 9KHz to 26GHz which won't be as useful.

[@johnyang799]

@@ejlockpix But that numbers you said don't mean it can measure low distortion. Can you measure ppm distortion? Like -120db 0.0001% thd? Also It's not necessary to have expensive audio precision to measure ppm distortion ( though it's perfect) I have focusrite forte. The linear reagion is around -18dbfs. So if I keep in that region I can measure 0.00005% thd.

[@johnyang799]

@@ejlockpix I thought you have audio precision or something like that. emm

[@johnyang799]

@@ejlockpix On the other hand. I think he said in earlier video that he only aim for about 0.003% for 1khz and 0.01% for 20khz. Decent and simple.

[@twotone3070]

Could a fellow youtuber help you out with the measurement, someone like Paul Carlson?

[@JohnAudioTech]

That may be an option. I really do need better equipment though.

[@saadaldossri5348]

Thanks

[@jasoneyes01]

I'm your enemy. We need more videos! It wouldn't be nice if I had to kidnap your kitty and hold him for ransom! John, can you produce a discrete push-pull headphone amplifier with some general purpose 2222 transistors or Jfets, diodes, and resistors. I'm struggling. I don't think you have a video on this topic yet. Sometimes I think you know more than Nelson Pass.

[@johnyang799]

kzbin.info/www/bejne/hXXZpHZ4n7-tb68

[@johnyang799]

But it's only half discrete. Thing is hard to design full discrete with excellent dc offset without dc servo. And within the voltage range of operation of headphones, opmaps seem to be much better. Add a high current cfa in the end making it composite amplifier will be very high performance. 0.00003% thd with 32ohm load.

[@milenedejong1400]

about the testing ... wouldn’t it be wonderfull to let someone else test it ? .. built the thing. stabelize it as much as you can and let someone else test it ... those numbers will speak for itself then..unbiased . maybe a small collaboration with mrcarlson or blueglow ?

[@JohnAudioTech]

Yes, I might have to do that.

[@russramirez5672]

John, consider purchasing a QuantAsylum QA401 Distortion Analyzer. These units are top notch and work very well. You may be able to find a used QA400, which is the unit I have, for a bit lower price. There are some related videos on the web page - quantasylum.com/products/qa401-audio-analyzer

[@JohnAudioTech]

Thanks for the link. It does appear deliver the goods for amp testing.

[@Larr.y]

Well, I didn't understand a single word of that ... I'm just here to catch up on Snickers!

[@bassblaster505]

the idea of the plate is EXACTLY like a guy named rcs16 used to make a replacement for the STK-0050 modules. has drivers and TO-3P outputs hanging off the sides onto a plate. audiokarma.org/forums/index.php?threads/stk-0050-replacement-for-sx-780-and-others.721181/ ALSO: i could even make a PCB design too. i work alot with EasyEDA

[@norullzz]

And an audio card that captures up to 92kHz with 192kS/s, that would get you >2x audible spectrum? eg shop.audioinjector.net/detail/Sound_Cards/Ultra+2

[@Elnufo]

Snickers!

[@Riker-ER]

My tuxedo cat would rather die than be confined to the indoors!

[@JohnAudioTech]

Cute kitty you have there! Cars, dogs, feral cats are too much of a danger around here.

[@Riker-ER]

@@JohnAudioTech Believe me, I understand! Mine survived thru prayers everyday! Still a lot of drama!

[@ВладимирСергеев-ц6ч]

Cat +++👍

[@fidziek]

1:30 - I hope your brain works

[@Mr_Meowingtons]

MEOW!!!

[@paulpaulzadeh6172]

Hi John, I have AP555 , you can send it to me I will measures everything for free , greetings from Sweden.

[@Dazzwidd]

Audible cat

[@Dazzwidd]

Discrete Cat

[@fatnindja]

John, for testing audio, I can recommend a very professional setup offered by a russian guy, who really busted a lot of myths about audio. His chanel is Stalker29218 kzbin.info/www/bejne/a6S7nIdqasqmg7s In few words, there is this program RMAA 5.5 (Right Mark Audio Analyzer). You need a quality sound card (external preferably) that supports 192khz/24bit recording and playback. The program plays a special audio test signal and then records the output of the test device through the input of the audio card. Then the program compares the 2 files and determines the most relevant measurements of audio equipment like frequency response, THD, Noise, THD+N, intermodulation, crosstalk etc. You can test by recording the output from a load resistor (through a voltage divider) or even directly test with dynamic load (speakers). The setup is easy but you need to have a reasonably decent sound card by reasonably decent I mean flat response and low noise. This is a waaay better way to measure these things. Let's be honest, your mp3 test signal is just not good enough for this task. Counting cells on the FFT by eye is not good enough. With this program you can also determine how bad is your mp3 player by playing the test signal on it, recording the output with your sound card and then compare the 2 files. RMAA will show you the results. This russian guy did a very extensive research and measurements of amplifiers, blind tests, audio myths etc. ... his channel may be an inspiration for you. It would be really cool if you switch to this way of measuring, so we can see the real qualities of your amp and other amps.

[@fatnindja]

He used some of your footage here: kzbin.info/www/bejne/g6O4koCFf9qSjLM :D