I decided to build myself an amp based on Cordell's circuit and have just completed a PCB design (on Christmas day!).
I tried to make a layout that's compact in size yet would offer a few interesting options. The circuit includes a TMC compensation, a baker clamp, a diamond buffer triple output stage employing up to 5 pairs of On-semi ThermalTrak devices, an op-amp based bias control, a solid state speaker relay, an output DC detection circuit that operates that solid state speaker relay, and an output over-current protection. I use through hole and SMT mixed design and managed to squeeze things into a double sided 230mm x 66mm area. Components are mostly on top side of the board, the output transistors and the Baker clamp components are the only parts installed on the bottom side.
The schematic and layout may appear to be a bit confusing as there seem to be a lot of redundant parts, such as paralleled zener diodes, doubled-up base stopper resistors, and doubled-up 0.1u bypass caps. These are for the SMT/through-hole part option purposes, just in case.
Not all component values have been calculated/simulated (I'll need to learn the LTspice, not even a newbie I am to simulation right now), but I supposed they wouldn't be too far off in terms of size of package/footprint which matters to the layout.
Before I order the PCB I'd very much appreciate the help of extra pairs of eyes just in case there are show-stoppers that I never would realized by myself until the boards come back. Thank your for any input.
Wish all a happy new year!
I tried to make a layout that's compact in size yet would offer a few interesting options. The circuit includes a TMC compensation, a baker clamp, a diamond buffer triple output stage employing up to 5 pairs of On-semi ThermalTrak devices, an op-amp based bias control, a solid state speaker relay, an output DC detection circuit that operates that solid state speaker relay, and an output over-current protection. I use through hole and SMT mixed design and managed to squeeze things into a double sided 230mm x 66mm area. Components are mostly on top side of the board, the output transistors and the Baker clamp components are the only parts installed on the bottom side.
The schematic and layout may appear to be a bit confusing as there seem to be a lot of redundant parts, such as paralleled zener diodes, doubled-up base stopper resistors, and doubled-up 0.1u bypass caps. These are for the SMT/through-hole part option purposes, just in case.
Not all component values have been calculated/simulated (I'll need to learn the LTspice, not even a newbie I am to simulation right now), but I supposed they wouldn't be too far off in terms of size of package/footprint which matters to the layout.
Before I order the PCB I'd very much appreciate the help of extra pairs of eyes just in case there are show-stoppers that I never would realized by myself until the boards come back. Thank your for any input.
Wish all a happy new year!
Hello,
There is a lot to review in this design. I actually received Bob's book from Santa & have not thoroughly read it, let alone absorb all the information.
I would suggest to perform the LTSPICE simulations, to perform virtual testing and analysis= sanity check.
I see that you have the RC Zobel R34/C86. L1 is to have a R in parallel? I guess that you will wrap L1 around a resistor in the layout, so that is why it is not shown on the schematic.
R47/48 are shown as 0? not sure why they are there, would think that they are optional.
Layout looks good, I do not see anything obvious = good job.
Good Luck & thanks for sharing
Rick
There is a lot to review in this design. I actually received Bob's book from Santa & have not thoroughly read it, let alone absorb all the information.
I would suggest to perform the LTSPICE simulations, to perform virtual testing and analysis= sanity check.
I see that you have the RC Zobel R34/C86. L1 is to have a R in parallel? I guess that you will wrap L1 around a resistor in the layout, so that is why it is not shown on the schematic.
R47/48 are shown as 0? not sure why they are there, would think that they are optional.
Layout looks good, I do not see anything obvious = good job.
Good Luck & thanks for sharing
Rick
Thanks, Rick. You must have done something good and Santa knew it well.
I Santa sent me two tubes of transistors 25x NJL3281 and 25x NJL1302
You're right about the simulation. I put off learning the LTspice for too long I should pick it up ASAP.
The missing paralell resistor on L1 was exactly what you thought. Good point. The layout software would not like a part in a schematic that does not have a place in the layout. I'll put a resistor into the schematic as a non electrical part, so that it does not upset the software while allowing the resistor to show up in the BOM.
R47/R48, along with R49/R50 and the 3k3 Rs off the 0.33R resistors, has to do with the behavior of the over current protection. I haven't go through the calculation so I put 0R for now.
Dug, obviously I'll go out and get a good helping-hand with a big glass for those LEDs. Thanks for you kind words.
I Santa sent me two tubes of transistors 25x NJL3281 and 25x NJL1302
You're right about the simulation. I put off learning the LTspice for too long I should pick it up ASAP.
The missing paralell resistor on L1 was exactly what you thought. Good point. The layout software would not like a part in a schematic that does not have a place in the layout. I'll put a resistor into the schematic as a non electrical part, so that it does not upset the software while allowing the resistor to show up in the BOM.
R47/R48, along with R49/R50 and the 3k3 Rs off the 0.33R resistors, has to do with the behavior of the over current protection. I haven't go through the calculation so I put 0R for now.
Dug, obviously I'll go out and get a good helping-hand with a big glass for those LEDs
. Thanks for you kind words.
LED A-C tester is easy, dug. And a tiny little dot from a pencil and the Cathode is marked good enough for larer mounting, don't You think?
nattawa, You got a Santa far better than mine. I just got a bag full of BC-capacitors.
R47/48 shorts the protectioncirquit BE on Q31 and Q32
They aren't to be mounted unless You want to run this one without protection, are You?
nattawa, You got a Santa far better than mine. I just got a bag full of BC-capacitors.
R47/48 shorts the protectioncirquit BE on Q31 and Q32
They aren't to be mounted unless You want to run this one without protection, are You?
Ok nattawa.
As I guessed then.
Well, otherwise the most looks good to me.
0.33Ohms emitter-resistors seems to be a little too much anyway.
Would let it go close to 4Amps /transistor before the protection should go in and do anything. This would be close to 0.15Ohm.
But as You say, you should do the math first.
Nice build this.
As I guessed then.
Well, otherwise the most looks good to me.
0.33Ohms emitter-resistors seems to be a little too much anyway.
Would let it go close to 4Amps /transistor before the protection should go in and do anything. This would be close to 0.15Ohm.
But as You say, you should do the math first.
Nice build this.
LED A-C tester is easy, dug. And a tiny little dot from a pencil and the Cathode is marked good enough for larer mounting, don't You think?
They are usually already marked. Just very small.
Hence the
@ gmphadte: The output stage has maximum of 5-pairs of power transistors, so it should be good for at least 300W into 8-ohm load. However, if such output power is intended, the power rails for the small current stages, including the input, VAS, and pre-driver, which is currently +/-60V should be re-sized accordingly, the TO-126 transistors used in the current design may have to be changed to higher voltage rating modles as well to suit the increeased supply rail voltage, KSA1381/KSC3503 for example.
Tweezers would work for some of the larger SMDs. I use a small flathead screwdriver, like one for eyeglasses or slightly larger to assemble tiny SMDs for prototypes. You dip out a glob of rosin onto a lid or something, then touch the screwdriver tip to the glob of rosin and the SMD part will stick to it. Tin one of the pads with solder. Touch the SMD part to the screwdriver tip so as you can properly place it on the pads, including the one you just tinned. Then melt the tinned pad to the SMD pin so it will hold in place. Then solder the rest of the pins in place. Non-prototype SMD circuits should be soldered by reflow or wave or something. I do SMD parts as small as 0201 size resistors, 0.6mmX0.3mm.
SOT-923, SOT-563 amd such is not much problem at all.
The schematic attached with the opening post has a wrong orientation. It is hardly a problem unless one views it with an ipad. So I uploaded the pdf files again. The schematic now has the correct orientation.
A slight change was made to the output sampling resistor R36. It now has a 1206 footprint, vs previous 0805, for a largest possible power rating SMD resistor for the available space, in order to reduce the distortion from the power consumption modulation to the resistor value, an effect discussed quite extensively in this forum. One may want to choose to use a thin film resistor for R36, as a thin film is not as much subject to such modulation as can be a thick film resistor. The through hole option for R36 remains unchanged at the same place. It allows the use of a 12.5mm lead spacing through hole resistor (or a stream of a few) when SMD is opted out for R36.
The prototyping of this amp may not be as easily going as may be a through hole PCB. Many SMD parts are surrounded by tall components and heat sink, demanding skills and patience when it come to troubleshooting, part changing, and signal probing. Proper soldering tools and skills will help here.
Cheers!
A slight change was made to the output sampling resistor R36. It now has a 1206 footprint, vs previous 0805, for a largest possible power rating SMD resistor for the available space, in order to reduce the distortion from the power consumption modulation to the resistor value, an effect discussed quite extensively in this forum. One may want to choose to use a thin film resistor for R36, as a thin film is not as much subject to such modulation as can be a thick film resistor. The through hole option for R36 remains unchanged at the same place. It allows the use of a 12.5mm lead spacing through hole resistor (or a stream of a few) when SMD is opted out for R36.
The prototyping of this amp may not be as easily going as may be a through hole PCB. Many SMD parts are surrounded by tall components and heat sink, demanding skills and patience when it come to troubleshooting, part changing, and signal probing. Proper soldering tools and skills will help here.
Cheers!
Last edited:
DC offset?
I see no provisions for an dc offset adjustment. No provision for a DC servo. Have you looked into this aspect of the design? What will be the offset voltage, what is your tolerance/limits? The input diff pair are discrete, thus matching maybe required to obtain the design goal.
How's the LTSPICE sim's going? Got your models?
I would assume that you have lots of experience with soldering SMT parts? I can help if you need some advise.
Current carrying capacity of the supply & output traces? are you planning on using 2oz copper plating? Leave some open solder mask in order to be able to deposit some extra solder to beef up the traces?
Not too sure about the optional cap underneath the SMT ecap's, such as C31/C59. Bottomside mounting of C31? will that not interfere? with the Heatsink? Not sure what side you will be mounting the O/P transistors.
Who will you get to fab the board, I can offer some sources.
Rick
I see no provisions for an dc offset adjustment. No provision for a DC servo. Have you looked into this aspect of the design? What will be the offset voltage, what is your tolerance/limits? The input diff pair are discrete, thus matching maybe required to obtain the design goal.
How's the LTSPICE sim's going? Got your models?
I would assume that you have lots of experience with soldering SMT parts? I can help if you need some advise.
Current carrying capacity of the supply & output traces? are you planning on using 2oz copper plating? Leave some open solder mask in order to be able to deposit some extra solder to beef up the traces?
Not too sure about the optional cap underneath the SMT ecap's, such as C31/C59. Bottomside mounting of C31? will that not interfere? with the Heatsink? Not sure what side you will be mounting the O/P transistors.
Who will you get to fab the board, I can offer some sources.
Rick
Good point about the DC offset, Rick. I personally would not like more than 10 or 20 mV DC at the output. So I'll use BL grade input transistor pair that has an Hfe in the range of 350-700. This should result in a +/-16mV DC output in a worst case scenario in terms of Hfe mismatching. A simple Vbe measurement/matching would perhaps be adequate then.
The amp was designed to have an input cap to block the DC at input, and the DC gain is unity. It could be possible to implement an DC servo, but it has to be on the bottom side of the board = hard to get at. So I opted it out for now. Perhaps an off-board DC servo could be built and tested then be made into the layout afterwards.
I have zero experience with LTSpice or any other simulation s/w I hope I can pick one up quickly. No simulation has been run on this design yet. I'd appreciate any fellow forum members who are interested in this design if they go ahead run and share the simulation results.
I'll be using 2-Oz copper though the layout does allow using home-made bus bar out of straight pieces of single strand copper wire on the power rails, their decoupling ground returns, and the output node, as a possible alternative when lighter weight copper PCB is used. These wires can be easily soldered on the bottom side of PCB.
As of the PCB trace current carrying capacity, All the high current traces are doubled up on both sides of the board. Power rail trace width totals 10mm each, o/p node trace totals 20mm wide. 10mm on 2-Oz copper is capable of 15A continuous current at 5C temperature rise, and 20mm on 2-Oz has about 7-mOhm resistance over a 100-mm run at 35C.
SMT caps placed within the SMT e-cap footprints are installed when SMT e-caps are opted out and through-hole e-caps are opted-in. All SMT component except 9 that the Baker clamp is made of are on the bottom side. The o/p transistors will be on the bottom side as well.
I have used a couple Chinese PCB shops in the past, cheap in price but the shipping would cost a fortune though they are usually ok in quality for the money for simple boards like this. I'll be glad to hear about your sources when the time comes. Thanks a lot.
The amp was designed to have an input cap to block the DC at input, and the DC gain is unity. It could be possible to implement an DC servo, but it has to be on the bottom side of the board = hard to get at. So I opted it out for now. Perhaps an off-board DC servo could be built and tested then be made into the layout afterwards.
I have zero experience with LTSpice or any other simulation s/w I hope I can pick one up quickly. No simulation has been run on this design yet. I'd appreciate any fellow forum members who are interested in this design if they go ahead run and share the simulation results.
I'll be using 2-Oz copper though the layout does allow using home-made bus bar out of straight pieces of single strand copper wire on the power rails, their decoupling ground returns, and the output node, as a possible alternative when lighter weight copper PCB is used. These wires can be easily soldered on the bottom side of PCB.
As of the PCB trace current carrying capacity, All the high current traces are doubled up on both sides of the board. Power rail trace width totals 10mm each, o/p node trace totals 20mm wide. 10mm on 2-Oz copper is capable of 15A continuous current at 5C temperature rise, and 20mm on 2-Oz has about 7-mOhm resistance over a 100-mm run at 35C.
SMT caps placed within the SMT e-cap footprints are installed when SMT e-caps are opted out and through-hole e-caps are opted-in. All SMT component except 9 that the Baker clamp is made of are on the bottom side. The o/p transistors will be on the bottom side as well.
I have used a couple Chinese PCB shops in the past, cheap in price but the shipping would cost a fortune though they are usually ok in quality for the money for simple boards like this. I'll be glad to hear about your sources when the time comes. Thanks a lot.
I was able to put in the layout a Toshiba HN3A51F dual PNP as an optional replacement to the discrete input transistor pair. The two closely matched transistors should save otherwise necessary DC offset trimming or manually matching the discrete transistors.
However, I had a hard time trying to locate a vendor that stocks HN3A51F. In fact few even has the model number listed at all. I'm wondering any Diyaudio members could share a source.
However, I had a hard time trying to locate a vendor that stocks HN3A51F. In fact few even has the model number listed at all. I'm wondering any Diyaudio members could share a source.
You need to connect grounds in a star ground configuration or inputs will be modulated by output signals.
Also keep the power supply smoothing as separate as possible from the amp to again stop ground modulation.
This is what I found from designing a mixer. My first pcb hummed very badly.
Once I followed the star ground rules there was no hint of hum.
Also, sometimes it is good to add decoupling between power amp output and driver/ltp stages.
Also keep the power supply smoothing as separate as possible from the amp to again stop ground modulation.
This is what I found from designing a mixer. My first pcb hummed very badly.
Once I followed the star ground rules there was no hint of hum.
Also, sometimes it is good to add decoupling between power amp output and driver/ltp stages.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.