Tcpip,
I have some cranky old software that doesn't like to produce images except in it's own format or Gerber files. I have also finally figured out how to make PDF files with it.
So tell me which format, PDF or Gerber you prefer. It may take me a couple of days to locate the old files so be patient.
By the way, they are double sided requiring soldering on the top unless you use though hole gromets. Also used use mini-resistors likew the Panasonic ERD or ERO sieries (body length is 3.5mm) because I was trying to squeeze into a smallish enclosure.
I have some cranky old software that doesn't like to produce images except in it's own format or Gerber files. I have also finally figured out how to make PDF files with it.
So tell me which format, PDF or Gerber you prefer. It may take me a couple of days to locate the old files so be patient.
By the way, they are double sided requiring soldering on the top unless you use though hole gromets. Also used use mini-resistors likew the Panasonic ERD or ERO sieries (body length is 3.5mm) because I was trying to squeeze into a smallish enclosure.
I now understand that it may not be of much use to me.... what I was hoping for was the raw data files in something like Eagle, so that I could edit them.sam9 said:
Thanks for your patience anyway. And please can you send me the PDF files, so that I can at least take a look at a layout other than Randy's, and it may help me make a layout of my own? Do you think it's feasible to fit one channel into a PCB of 3.2"x4" size? (That's the size that the free version of Eagle allows.) I don't mind DSPTH, there are fabs here who do a very good job with them at affordable rates.Thanks again.
"Do you think it's feasible to fit one channel into a PCB of 3.2"x4" size?"
I think so if you look for items that can be placed somewhere off the PCB. For startes: fuses, output inductor and damping resistor, Zobel network.
Some things can be made smaller. A few examples: Use a 47uF bi-polar cap for the input instead of oppossed tantalums. Use single pins rather than terminal blocks or just solder a wire from a single hole - these don'y need to be at the board's edge which could ease the layout constraints. Stand the 5W resitors on end, but take care because they are subject to stress and the solder joints are easily damaged.
You may be able to make a more compact layout by connecting the Vbe (bias) transistor to the board bt wires rather than placeing the footprint right next to the heatsink/board-edge.
I think so if you look for items that can be placed somewhere off the PCB. For startes: fuses, output inductor and damping resistor, Zobel network.
Some things can be made smaller. A few examples: Use a 47uF bi-polar cap for the input instead of oppossed tantalums. Use single pins rather than terminal blocks or just solder a wire from a single hole - these don'y need to be at the board's edge which could ease the layout constraints. Stand the 5W resitors on end, but take care because they are subject to stress and the solder joints are easily damaged.
You may be able to make a more compact layout by connecting the Vbe (bias) transistor to the board bt wires rather than placeing the footprint right next to the heatsink/board-edge.
Thanks for these inputs. I'll try now. I didn't understand that last bit about the Vbe transistor's placement, though. And do you really recommend a bipolar cap for carrying the signal? What kind of bipolar is this? Won't it be better to just use two 10uF MKP/MKT caps in parallel?sam9 said:
BTW, I got a diyaudio-routed message from you. I tried replying to it, but it immediately bounced. It appears that your concentric dot net email address is not valid any more. Is that the address you've retained in your diyaudio profile? Can you write a message to me giving me any alternate email address?
Thanks and regards,
Tarun
I'll fix my member profile so you can email me. I've changed my ISP.
"And do you really recommend a bipolar cap for carrying the signal? What kind of bipolar is this? Won't it be better to just use two 10uF MKP/MKT caps in parallel?"
Yes, a bipolar is sub-optimal, but it takes up a lot less space (8mm diameter) which seemed to be an issue. You can find them in the online digikey catalog - they are made by Panasonic they just eseentially two caps constructed in oppossed orientation in a single can. Two 10uF film caps in parrallel would be superior and just one would do the job. The deal with my bipolar suggestion is that an electrolytic of sufficiently high value has a such low resistance that the signal pases through with out charging or discharging it -thus is in almost not there. A 47uF bipolar electrolytic is actually like two 100uF caps. I would not do this so long as I had the board space available for a 10uF film cap.
The Vbe comment is based on experience laying the PBC. It is electrically connected pretty intemately with the VAS but you want to have it physically near either the heatsink and output devices or near one of the small driver transistors in order for it to be in contact with then for thermal tracking. It makes routing the traces awke=ward some times because the best placement electrically is not compatable with the best placement thermally. So if you are trying to layout a small PCB this is one more problem to deal with. Take a look at how one guy solves this: http://www.aksaonline.com/products/products_aksa55.html. Notice that the three wires don't even come of the board near each other - this is very cool solution. Obvious and simple once someone else thinks of it. (It's an EF rather than CFB design but the idea would work for both.)
"And do you really recommend a bipolar cap for carrying the signal? What kind of bipolar is this? Won't it be better to just use two 10uF MKP/MKT caps in parallel?"
Yes, a bipolar is sub-optimal, but it takes up a lot less space (8mm diameter) which seemed to be an issue. You can find them in the online digikey catalog - they are made by Panasonic they just eseentially two caps constructed in oppossed orientation in a single can. Two 10uF film caps in parrallel would be superior and just one would do the job. The deal with my bipolar suggestion is that an electrolytic of sufficiently high value has a such low resistance that the signal pases through with out charging or discharging it -thus is in almost not there. A 47uF bipolar electrolytic is actually like two 100uF caps. I would not do this so long as I had the board space available for a 10uF film cap.
The Vbe comment is based on experience laying the PBC. It is electrically connected pretty intemately with the VAS but you want to have it physically near either the heatsink and output devices or near one of the small driver transistors in order for it to be in contact with then for thermal tracking. It makes routing the traces awke=ward some times because the best placement electrically is not compatable with the best placement thermally. So if you are trying to layout a small PCB this is one more problem to deal with. Take a look at how one guy solves this: http://www.aksaonline.com/products/products_aksa55.html. Notice that the three wires don't even come of the board near each other - this is very cool solution. Obvious and simple once someone else thinks of it. (It's an EF rather than CFB design but the idea would work for both.)
tcpip said:I'm not sure I understood this correctly, so please can you help me? I was wondering what if I connect eight signal diodes in two series-chains of four each, between input and signal-ground. Each chain of 1N4148 will have four diodes, hence will conduct only when the potential difference will go beyond about 2.4V. Will such diodes also introduce distortion?
And what about the 1N4007 diodes that are used between the speaker output and the supply rails of the power amp? Will those also cause distortion?
I'm really new to all this, so some of my questions may sound stupid. Sorry.
The diode strings will probably work, and with negligible distortion.
The Hitachi MOSFETs have built in diodes, so the 1N4007s are not needed. However, this is what they are intended to do: protect the output transistors from inductive spikes. If you were to wiggle the speaker cables while delivering high output (causing them to make intermittent connections) the inductance of the speaker system would send high voltage spikes back at the amp outputs, possibly causing them to exceed their breakdown rating and short out destructively. Those 1N4007 diodes should never conduct, unless there is a highly inductive load like a typical speaker, and rapidly changing waveform. The diodes are clamps on the output voltage, so it can't spike above (or below) the positive (negative) supply voltage, damaging output transistors.
The diodes are a vestigial holdover from the bipolar version of this amp, where they most definitely are needed since BJTs don't have an intrinsic clamp diode. Like most engineers, Slone leverages known good designs wherever possible.
Not only are the diodes a vestigial remnant of bipolar amps,
they are an artifact created by short protection circuits -
an amp without foldback current limiting is not going to see
enough flyback to break the reverse bias voltage of the BE
diode.
I put them in anyway, 'cause I wear a belt and suspenders.
they are an artifact created by short protection circuits -
an amp without foldback current limiting is not going to see
enough flyback to break the reverse bias voltage of the BE
diode.
I put them in anyway, 'cause I wear a belt and suspenders.
Thanks for the answers. Makes sense.sam9 said:
That photo really made the whole thing fall into place. Yes, it's a very cool design. I guess Hugh Dean is a cool designer.
Seeing the size of his AKSA55 PCB, I now think perhaps I can try doing a "Slone-Fig11.4" PCB within the bounds of Eagle. Or else I'll just edit the layout you did in Vutrax. I'm waiting for it eagerly.
Tarun
I'm very interested in building one of the CF-MOSFET amps that are shown on the Sealelectronics site. Any more information would be appreciated, like what exactly comes with the kit. Are power supplies included? Binding posts? Is this a one channel or stereo amp? Where would one get an enclosure? Etc.
Midficollege said:
1 Kit includes PCB, heatsink, components for one channel, assembly instructions. You need 2 kits for stereo, obviously
You can buy PS filter capacitors from Mr Slone, but you'll additionally need (3rd party sourcing): transformer(s) - they're heavy and shipping costs cut in here so get them from a nearby shop, binding posts, RCA connectors and connecting wires. You might also want a case to put it in, unless you're like me and like it naked tcpip said:
The original value is two series 47uF tantalums which comes out to ~20uF, so it would be nice to use MKPs here. However nothing beats a piece of wire instead of the cap, as unity gain at DC is assured by the cap in the feedback loop. Unless there's any chance to put 5VDC or more at the input I'd leave those out.
If you decide to use MKP then try to get lower voltage ones to save some space... 10uF/100V is kinda huge
DC coupled input has a certain logic but I remain nervous about it, nonetheless. Reliance on just the NFB return to protect you takes more faith than I've got.
On the other hand preamps often have output caps and it seems to me that having a cap at thr preamp output and another at the amp input is redundant. In the commercial world it makes sense as you don't know what you customer is going to do. For diy where you know (or think you know ) the environment in which each piece of gear is operating, you can safely eliminate some redundancies. If I had built both the preamp and the amp or was building an integrated amp, direct coupling the stages would not bother me.
I would also want to be sure that both were plugged into the same mains circuit to be sure they saw an identical grond potential. If there were a difference in potential seen by each unit that would cause a DC offset that is as real as connecting a battery to the amp's input.
On the other hand preamps often have output caps and it seems to me that having a cap at thr preamp output and another at the amp input is redundant. In the commercial world it makes sense as you don't know what you customer is going to do. For diy where you know (or think you know
) the environment in which each piece of gear is operating, you can safely eliminate some redundancies. If I had built both the preamp and the amp or was building an integrated amp, direct coupling the stages would not bother me. I would also want to be sure that both were plugged into the same mains circuit to be sure they saw an identical grond potential. If there were a difference in potential seen by each unit that would cause a DC offset that is as real as connecting a battery to the amp's input.
And Randy says that if you really want to go first-class, you should use a 250V MKP cap at the input, to protect the rest of the circuit against accidental high-voltage DC. He recommends this not for sonic reasons, but for protection.lucpes said:If you decide to use MKP then try to get lower voltage ones to save some space... 10uF/100V is kinda huge
I guess you've seen the sale of GE MKP caps on Madisound.com? They're at really low prices.
Appearently he built quite a few custom amps for musicians. The input connectors are often TRS (1/4" phono) jacks. I've seen a few roadies in my time (the Greatfull Dead used to make appearences near here a couple times a year) some know what they are doing and some sholudn't be allowed to use sissors without supervision even when not "chemically enhanced". Thus I imagine Slone recognizes real need to protect his handiwork from having all kind of thing plugged into the input.
Regarding input caps, I suggested lower voltage MKPs because of the space issues that were previously discussed.
For personal use it's ok to go capless , if you're willing to take the risk (or keep only one cap either at the preamp out or amp in if necessary) Sam9 is right about 'plugging' things into the input... you'll never know what will come in there when talking about commercial units.
For personal use it's ok to go capless , if you're willing to take the risk (or keep only one cap either at the preamp out or amp in if necessary)
Sam9 is right about 'plugging' things into the input... you'll never know what will come in there when talking about commercial units.I guess this bit of Randy's personal life is not known widely, but he plays some musical instruments (I forget which), and he's been part of a travelling band. He loves music, and listens quite a bit, and I believe he still plays for his own pleasure. So I guess he knows very well what the "pro audio" environment is like, at least for performing tours.
Sometimes, audio designers who believe in measurement and objective tests, double-blind comparisons, etc, are branded as dry and lacking the urge to sit and listen. Both Randy and Seigfried Linkwitz are music lovers, though both of them have an engineering approach to audio system design and evaluation. I was told SL likes attending live concerts. Both these people are my inspirations for audio system design and evaluation.
Sometimes, audio designers who believe in measurement and objective tests, double-blind comparisons, etc, are branded as dry and lacking the urge to sit and listen. Both Randy and Seigfried Linkwitz are music lovers, though both of them have an engineering approach to audio system design and evaluation. I was told SL likes attending live concerts. Both these people are my inspirations for audio system design and evaluation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.