Mikek,
I have experience of Self's circuit, although I did modify it somewhat. Here's my eperience:
1. Replace diff pair feed with a resistor, giving 1mA stage current.
2. Replace the 2K0 collector resistor on T1 with 1K5.
3. Power with +/-15V rails.
4. Ramp up the current through the output stage to 12mA.
5. Delete the emitter follower output stage, taking global feedback from the collector of the VAS/CCS.
This gives a Zout of around 35R, with excellent stability and FR 1dB points at around 15Khz and 50KHz. These are real world figures. The input devices are BC557, VAS is BC550.
Cdom is about right, perhaps a tad higher (18pF).
It sounds VERY good, and I don't see that anyone should be criticizing Doug Self, who has done a great service, albeit imperfect at times, to the DIY and pro-audio crowd. I am a little concerned at the reluctance to break out the iron and test these circuit blocks in environments other than PSpice.
Cheers,
Hugh
I have experience of Self's circuit, although I did modify it somewhat. Here's my eperience:
1. Replace diff pair feed with a resistor, giving 1mA stage current.
2. Replace the 2K0 collector resistor on T1 with 1K5.
3. Power with +/-15V rails.
4. Ramp up the current through the output stage to 12mA.
5. Delete the emitter follower output stage, taking global feedback from the collector of the VAS/CCS.
This gives a Zout of around 35R, with excellent stability and FR 1dB points at around 15Khz and 50KHz. These are real world figures. The input devices are BC557, VAS is BC550.
Cdom is about right, perhaps a tad higher (18pF).
It sounds VERY good, and I don't see that anyone should be criticizing Doug Self, who has done a great service, albeit imperfect at times, to the DIY and pro-audio crowd. I am a little concerned at the reluctance to break out the iron and test these circuit blocks in environments other than PSpice.
Cheers,
Hugh
peranders said:
Hi peranders
Are those FR4 boards on your website?
I would like to know what schematic-capture and layout package you use for your work...cheers.
P.S: don't trust autorouters...don't like the loss of control...
....wouldn't mind one that i could overide whenever i feel like...at any point in its cycle...
PCB program
Hi mikek,
I use Eagle, it's cheap and there is a free version to try it out
I have also worked with PADS power logic and PADS power PCB. They are excelent programs, but very expensive
I have not yet tried any autorouters that worked, but please tell me what other programs are good.
For simulation I use ORCAD PSPICE, but it's a real drag having to keep 2 different versions og schmatics updated i different programs.
\Jens
Hi mikek,
I use Eagle, it's cheap and there is a free version to try it out
I have also worked with PADS power logic and PADS power PCB. They are excelent programs, but very expensive
I have not yet tried any autorouters that worked, but please tell me what other programs are good.
For simulation I use ORCAD PSPICE, but it's a real drag having to keep 2 different versions og schmatics updated i different programs.
\Jens
Mikek,
I wasn't really having a shot at you, though God knows you are a *****ly, noisy bugger and it wouldn't be difficult......
And yes, I didn't read all your posts. Time forbids. Do you think I should???
I do agree over Per's boards. Simply magnificent, a labour of love. But oh, such a high component count!
Cheers,
Hugh
I wasn't really having a shot at you, though God knows you are a *****ly, noisy bugger and it wouldn't be difficult......
And yes, I didn't read all your posts. Time forbids. Do you think I should???
I do agree over Per's boards. Simply magnificent, a labour of love. But oh, such a high component count!
Cheers,
Hugh
Re: PCB program
Multisim professional suite, or power professional suite:
www.electronicsworkbench.com
JensRasmussen said:
Multisim professional suite, or power professional suite:
www.electronicsworkbench.com
Just a generality. I have built circuits identicle to each other but with different PCB layouts. One would oscillate if you even look at it funny, the other was stable under deliberately adverse conditions.
The point is that two people can build the same circuit and get differeng results. There is a lot of info around about circuits per se, but I've yet to see anyone publish really comprehensive advice on laying out pcbs. Therte are rules of thumb, suggestions, guidelines etc - but there remains a lot of trial and error. I've come to accept that I will layout and build things ***at least*** 3-4 times before I'm satisfied with it. (or even before it works right!) Some people are better at this but to me it is more like Rubik's cube.
The point is that two people can build the same circuit and get differeng results. There is a lot of info around about circuits per se, but I've yet to see anyone publish really comprehensive advice on laying out pcbs. Therte are rules of thumb, suggestions, guidelines etc - but there remains a lot of trial and error. I've come to accept that I will layout and build things ***at least*** 3-4 times before I'm satisfied with it. (or even before it works right!) Some people are better at this but to me it is more like Rubik's cube.
Dear MikeK:
>THD in simulation that gives WORSE results than those ascribed to a circuit in prototype, constitutes grounds for the gravest suspicion.<
This is not meant as a comment on Self's circuit, but I have designed circuits that performed considerably better in real life than they did on the simulator, including distortion spectra. As has been said many times in the past, this is probably a problem with the device models rather than the simulator engine. I frequently use Japanese semi's, and reasonable models for these are very hard to come by.
In your copy of MultiSim Pro (version 7?), did you get the Model Maker? If so, did you generate a spice model in Model Maker and compare the results with some of the spice models (for the same device) that you can get publically from, say, Orcad? What kind of differences did you then see in the simulation results?
regards, jonathan carr
>THD in simulation that gives WORSE results than those ascribed to a circuit in prototype, constitutes grounds for the gravest suspicion.<
This is not meant as a comment on Self's circuit, but I have designed circuits that performed considerably better in real life than they did on the simulator, including distortion spectra. As has been said many times in the past, this is probably a problem with the device models rather than the simulator engine. I frequently use Japanese semi's, and reasonable models for these are very hard to come by.
In your copy of MultiSim Pro (version 7?), did you get the Model Maker? If so, did you generate a spice model in Model Maker and compare the results with some of the spice models (for the same device) that you can get publically from, say, Orcad? What kind of differences did you then see in the simulation results?
regards, jonathan carr
jcarr said:
Hi Jonathan,
Thanks for info. Yes....i've previously generated spice models for Zetex medium power devices in model maker.....the results indicate that Zetex have done a pretty good job with their models in respect of consistency....
....I couldn't actually obtain THD results either in simulation, or in practice without modifying Self's frequency compensation, (1K in series with 15p, across the feedback resistor),...
... I found that rather disturbing...
Place parts like the schematics (if it's draw by any sense)
PCB layouts rarely match parts placement on the printed circuit board. I wonder if this is fact a detriment to the schematic capture phase of design. Good schematics are drawn with conventions for the easiest understanding of circuit function. Most schematics are drawn with inputs on the left and outputs on the right. Positive power supply connections are usually show at the top of the page and negative supplies at the bottom. Right and left channels are often on separate pages or as single schematic for both with a note on Reference designator differences between channels. Power supplies, protection circuits, and other common circuits are often separate sections or pages on the schematic. The document describes the function of the circuit for design documentation and troubleshooting. It is not a map of the physical location of the parts on a PCB and is not a map of there location. When a device is built the assembly house doesn't use the schematic but works from an assembly drawing and the silk-screen on the PCB.
The physical placement will be determined by the location of input and output jacks, chassis size, component size and height (an overlooked factor in the two dimensional representations that has tripped many up in layout), and by the number and shape of the PCBs. The step of parts placement is critical and is the necessary transition stage between schematic and the constraints of the PCB layout. It is the step where critical signal path lengths, decoupling, grounding, crosstalk, high current and small signal interaction are considered. It is where a good PCB layout and the actual physical constraints of the circuit in the real world are considered. To anticipate these factors at the schematic capture phase is practically impossible. It would likely result in a would result in schematic that would be confusing in terms of quickly communicating the circuit function to somebody else. The schematic is not a complete representation circuit of the physical circuit and jumping from the schematic to the layout without careful consideration of this will results in a compromised schematic and a very compromised PCB layout. Go reverse engineer a good PCB layout to derive the schematic and see if you don't wind up with a pretty confusing schematic until you redraw it, even when drawing it in CAD program. In fact I just did this a couple of weeks ago for some amp boards from a commercial amp that Jock sold me.............
PCB layouts rarely match parts placement on the printed circuit board. I wonder if this is fact a detriment to the schematic capture phase of design. Good schematics are drawn with conventions for the easiest understanding of circuit function. Most schematics are drawn with inputs on the left and outputs on the right. Positive power supply connections are usually show at the top of the page and negative supplies at the bottom. Right and left channels are often on separate pages or as single schematic for both with a note on Reference designator differences between channels. Power supplies, protection circuits, and other common circuits are often separate sections or pages on the schematic. The document describes the function of the circuit for design documentation and troubleshooting. It is not a map of the physical location of the parts on a PCB and is not a map of there location. When a device is built the assembly house doesn't use the schematic but works from an assembly drawing and the silk-screen on the PCB.
The physical placement will be determined by the location of input and output jacks, chassis size, component size and height (an overlooked factor in the two dimensional representations that has tripped many up in layout), and by the number and shape of the PCBs. The step of parts placement is critical and is the necessary transition stage between schematic and the constraints of the PCB layout. It is the step where critical signal path lengths, decoupling, grounding, crosstalk, high current and small signal interaction are considered. It is where a good PCB layout and the actual physical constraints of the circuit in the real world are considered. To anticipate these factors at the schematic capture phase is practically impossible. It would likely result in a would result in schematic that would be confusing in terms of quickly communicating the circuit function to somebody else. The schematic is not a complete representation circuit of the physical circuit and jumping from the schematic to the layout without careful consideration of this will results in a compromised schematic and a very compromised PCB layout. Go reverse engineer a good PCB layout to derive the schematic and see if you don't wind up with a pretty confusing schematic until you redraw it, even when drawing it in CAD program. In fact I just did this a couple of weeks ago for some amp boards from a commercial amp that Jock sold me.............
Hi Fred,
I agree and would like to add a couple of comments.
Don't be afraid to revise the schematic based on how the layout is turning out. Especially when you are using dual or quad op-amps.
Another trick that can help prevent problems with grounding when using an auto-router is to electrically separate the digital and analog ground planes in the schematic and bridge them at the star ground point with a zero-ohm resistor. At layout time, replace the zero ohm resistor with a copper pour. Also be sure to rout any signals that have to pass between the analog and digital sections over the top of that ground plane pour.
Phil
I agree and would like to add a couple of comments.
Don't be afraid to revise the schematic based on how the layout is turning out. Especially when you are using dual or quad op-amps.
Another trick that can help prevent problems with grounding when using an auto-router is to electrically separate the digital and analog ground planes in the schematic and bridge them at the star ground point with a zero-ohm resistor. At layout time, replace the zero ohm resistor with a copper pour. Also be sure to rout any signals that have to pass between the analog and digital sections over the top of that ground plane pour.
Phil
Fred's observations about physical constraints, i.e., power, ground and I/O connection, enclosure dimensions and so forth.
As an example, I recently laid out a moderate power amp where I could use on-board heatsinks. This removed the physical contraint of having to line up the output devces at the board edge. It was really interesting, how much cleaner the layout was when this flexability was added. It was much easier to keep the input and VAS networks away from the output section. On another ocassion, I realised that a SF Mosfet OPS eliminated the need for thermal tracking - the result was tha even though in simulation the MOS version of this project was the noisiest, when actually constructedit came out nearly equal or better than a couple of BJT versions I was considering.
As an example, I recently laid out a moderate power amp where I could use on-board heatsinks. This removed the physical contraint of having to line up the output devces at the board edge. It was really interesting, how much cleaner the layout was when this flexability was added. It was much easier to keep the input and VAS networks away from the output section. On another ocassion, I realised that a SF Mosfet OPS eliminated the need for thermal tracking - the result was tha even though in simulation the MOS version of this project was the noisiest, when actually constructedit came out nearly equal or better than a couple of BJT versions I was considering.
If I am not mistaken.......
Self says "typical" op-amp, not "actual op-amp that I built that doesn't work".
So.....if feedback can fix anything, then you should be able to make anything with 2N2222s and 2N2102s, right?????? No sense for all those companies to make all those other transistors. More marketing rubbish, I guess...........
Jocko
Self says "typical" op-amp, not "actual op-amp that I built that doesn't work".
So.....if feedback can fix anything, then you should be able to make anything with 2N2222s and 2N2102s, right?????? No sense for all those companies to make all those other transistors. More marketing rubbish, I guess...........
Jocko
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.