Well, back again...sorry about that and thanks for the encouragement, Carlo!
The power supply, as I mentioned, works exactly as the schematic shows and I was able to draw up a pcb on a trial version of Protel 99SE in about an hour. I think that I could have squeezed things a little tighter, but I had a 6" X 4.5" board so I gave components plenty of room and drew extremely wide copper traces, bottom side only. I used 4 medium sized electrolytic Panasonic 6800 uF 63 volt rated TSH series caps in place of KG's 4 4700 uF caps. I had a little trouble squeezing in the horizontally mounted 20k pot trimmers to allow access afterwards with a small screwdriver so you should place the pots so the turn-screw faces outward, or something like that....
With regards to the LM317/337's, it is advisable to mount them to a heatsink using proper plastic standoff collars (?) or else they will definitely pass on current to the chassis. How do I know this? Hmmnn.......bzzt..
On the bottom right of the PSU schematic, KG has drawn a lead with one end connected to the - in of the OPA548 and the other end going out to a 10K resistor and then to "+16.4 Vdc". Call me stupid, but it took me a bit to realize that this actually connects back to the +16 Vdc output leg of the positive rail.
It is especially important to keep in mind that on each OPA548, pin #3 and pin #4 should both be connected to the -24Vdc supply rail and pin #5 to the +24Vdc supply rail. I ran wires from just after the last filter capacitors on the negative and positive rails (which should be +24Vdc and -24Vdc at that point) and connected them to the respective pins on the OPA548's. You must do this for each OPA548.
The cost for the power supply was also reasonable; the 6800 uF caps cost about $7 ea. from Digikey, the transformers cost about $10 ea. at Radio Shack (although you might want to invest in a higher quality unit!) and the 20K pots (Digikey # 3009P-203-ND) were $2.72 ea. You may want to get beefier trim pots yourself, but these seem to work okay... The most expensive pieces were the OPA548's (Digikey part # OPA548T) at $10.48 ea. and the Ref02 (Digikey #REF02AP-ND) at $3.56.
In terms of resistors, I ordered a bunch of 1 % metal film Yageo resistors from Digikey as well. They cost something crazy-cheap like .018 cents each so I was able to get 20 in each value and hand match them pretty easily. About 25% of the resistors used were Yageos and the rest were Dale/Vishays from Welborne Labs. They cost a lot more--$.50 each--but their values were noticeably more accurate and they fit like a charm into KG's pcb.
I placed the PSU into a separate 19" chassis with 3 external cable leads (+16.40 Vdc/-16.40 Vdc and ground) to allow me to hook up different amps to the unit and in that respect, I feel that the "ultra-regulation" that KG refers to is really worth it since you will end up with a pretty versatile, high performance power supply that'll last you a lifetime or 'till the warranty runs out...
On the other hand, I've been listening to the Gilmore amp and the Gilmore power supply for a few days now and I can't really tell much difference between the amp running with the final stage of regulation and with it running only with the LM317/337 stage. The amp seems more impervious to power supply changes than the Szekeres, that's for sure.
Updated 9/12/02 !!!!
The Gilmore Power Supply PCB resized for smaller enclosures and surface mount Talema 70075 transformer
Here is a new pcb for the Gilmore Power Supply that measures 8" x 3.8".
The board is designed to accept the PCB mounted Talema 70075 series toroidal transformer or any other 72 mm x 72 mm surface mount transformer with similar pinouts.
Click here for more information on Talema Transformers
You can download the re- laid out Gilmore pcb design here:
The pcbs should be printed out at 8.0" x 3.8" at 300 dpi.
High Resolution Gilmore PSU 8.0" x 3.8 " bottom layer .psd file
Lower resolution Gilmore PSU 8.0" x 3.8" bottom layer .gif file
Composite Master Picture
Drill Drawing 8000 x 3800 mil
Top Overlay with Pads
Top Silkscreen
*note* I tried to get rid of as many extra wires as possible, but if you decide to try this design, you'll need to solder an insulated wire from the pad labeled "+24Vdc out" to the pad labeled "+24Vdc in". Please refer to the above listed "Composite Master" to verify this connection.
Updated 09/14/02
The Gilmore 6.25" x 4.5" Power Supply PCB:
Here is a newer version of the power supply PCB I used in my original amp made from Kevin Gilmore's schematic.
Gilmore Power Supply 6.25" x 4.5" PCB -- High Resolution .psd File -- please download and save with MS Internet Explorer or use "Save Link Target as..." command with Netscape!
Lower resolution Gilmore PSU 6.25" x 4.5"bottom layer .gif file
Gilmore 6.25" x 4.5" Power Supply Composite Master Pic. with notes
Gilmore 6.25" x 4.5" Power Supply Composite Clean with no notes
Gilmore 6.25" x 4.5" Power Supply Top Overlay
Gilmore 6.25" x 4.5" Power Supply PCB Drill Guide
UPDATED 02/26/03
FRED/Schottky Gilmore 6" x 9" High End Power Supply PCB Design :
You can download the FRED/Schottky Gilmore PSU files by clicking on the links below:
FRED Gilmore Bottom .PSD 520K
FRED Gilmore Bottom .GIF 83K
FRED Gilmore Composite w/ Pads 99K
FRED Gilmore Components 30K
FRED Gilmore Master Composite 125K
FRED Gilmore Pads 18K
Here's a picture of an earlier version of the board without on-board rectifiers (they were placed on another board) but it should give you a rough idea of what a finished PSU board looks like.
Please note that the design is made to fit on a 6" x 9" board (I used the Injectorall 1 sided 6x9 copper clad board-Digikey # PC11-ND). There is enough space for up to six 10,000uF capacitors and/or any combination of values that will fit on the board. There is also enough space for large sized polypropelene capacitors both before the OPA548 and after.
The Bottom artwork MUST be printed out at 6" x 9"!!
I left enough space for anyone who wants to use heat sinks with their TO-220 form factor FREDS. Please note that you will need 8 FREDs to complete the 2 full wave bridges.
Also, it is highly recommended that you heat sink the particular FRED diode leg you are soldering due to the possibility of heat damage to the diode.
I also mounted the LM317/337's on "rails" in order to accomodate a variety of different size heat sinks. You can now move the LM317/337's back and forth about 2 to 2.5 inches to mount much larger heat sinks.
*NOTES*
First of all, as all of us budding DIY'ers know, working with mains/ac electricity can kill us/you/me/anyone. It is very dangerous and all necessary precautions should be taken to ensure that you don't hurt yourself or anyone else. If you feel uncomfortable at the thought of working with fairly high voltages, please do not try to build this power supply on your own!
That said, this version of the Gilmore Power Supply PCB features the REF02 and OPA548 voltage regulators. If you look at the schematic, you'll see that you need (1) REF02 and (2) OPA548 chips to complete the PSU properly. In addition, you will also need to get (1) LM317 and (1) LM337 voltage regulators for the intial regulation stage.
I tried to get rid of as many extra wires as possible, but if you decide to try this design, you'll need to solder an insulated wire from the pad labeled "+24Vdc out" to the pad labeled "+24Vdc in". Please refer to the above listed "Composite Master" to verify this connection.
Furthermore, when constructing the amp, please be aware that the LM317 and the LM337 are mounted facing each other. There is some room for a small/medium board-mounted heatsink, but if the temp. gets too hot, you can always run some lead wires from the "Adj", "Vin" and "Vout" solder points on the PCB to chassis mounted heatsinks.
Next, please note that the 2 OPA548s are facing away from each other. Again, there's a lttle room on the board itself to mount the OPA548s and their respective heatsinks, but you may want to consider larger chassis mounted heatsinks for additional temperature control.
You should also note that only the bottom 2 pins on the (2) 20K trim pots are actually connected to anything. The top pin (the one closest to the rotating screw or dial) can either be cut off or used as a mechanical anchor for the trim pot--as long as you make sure it doesn't touch any live circuit paths.
The 5uF capacitors, as shown in KG's schematic, are optional. There is some space on the board to place the 4 capacitors (depending on their size), but, as Kevin says, these caps are not absolutely necessary.
As for the main filter capacitors, I marked them as "6800 uF" caps simply because that's what I used in mine. You can fit smaller radial-type caps or larger ones, too. I tried to leave some extra space for 10,000 uF caps but I don't know for sure if any will actually fit or not. I ended up using (4) 6800uF Panasonic TSH caps and they are actually a bit smaller than the space marked on the board.
As I mentioned earlier, I'm also using two Radio Shack 25 volt/2 amp (?) transformers that cost about $12 each. They actually do a pretty fine job, although they do get kind of hot and I don't know how well they'll survive summer temperatures.
Finally, for all my fellow newbies, please remember to use proper standoffs to lift the board away from any metal surfaces that could short the power supply and surprise the heck out of you.
Good luck and stay safe...
Making the PCB:
In terms of making the PCB's, I'm using a fairly crude, but pretty effective laser printer/transparency/ironing board combination that works okay with minimal investment in materials. First, print out the pcb pattern on a plastic transparency (3M Transparency Film for Laser printers or any similar product).
* The power supply board transparency should be printed out at 6.25" x 4.5" or as close to these dimensions as possible. (actually, 6.250 x 4.503 at 300 dpi)* I sized it to fit on a standard Radio Shack copper board that costs about $3.99 US.
Next, sand and polish the appropriately sized copper board with extra, ultra fine grain sandpaper.
**Wipe the surface of the copper board carefully with a dry cloth before
putting the transparency on top in order to ensure a clean bond between the copper and the ink transfer**
Place the transparency ink side down on top of the copper board.
Next, take a regular household iron and turn it up to max. temp. Wait for the iron to heat up then carefully run it over the plastic sheet for a minute or two until the ink transfers completely/almost completely to the copper.
Please check out Jeff Noxon's site for more detailed information on DIY PCB's
Here's a link to an Excellent Site on How to "Press and Peel" Your Own PCB's.
Here's a link to the Techniks "Blue Transfer Film" Press 'n Peel Site
*TIP*
As you slowly peel of the transfer, you may find that some lines on the transparency may not have transferred completely to the copper board. In this situation, cover the board again with the transparency and using the tip of the iron, re-iron the partially transferred print. Using this method-- slowly peeling, inspecting and then re-ironing with the tip, you should be able to get close to a 100% transfer rate. However, if you take the transfer off completely from the copper board, you will have a difficult time re-registering the transfer exactly. The best technique is to do this slowly--take your time, peel off only a bit at a time and just re-iron the print that didn't transfer properly...
(In any case, its a good idea to cover up any "thin" areas --where you can see some copper showing under the toner ink-- with a black Sharpie. You should, as a precaution, also re-trace any thin lines with a fine tip Sharpie.)
The nice thing is that if you screw up, you can erase the partially transferred ink pattern with some acetone and start over again. Pretty cheap! It is extremely important that you inspect the transfer at this point! If there are any breaks in the pattern or flaws, you can usually touch it up with a fine point Sharpie. If there are unexpected breaks in your transfer, the circuit won't work. You can always try to solder over a broken connection afterwards, but its a pain to do this. Catching mistakes at this point will save you a lot of time.
Once you've inspected the transfer pattern on the copper board, you can then dump it in a bath of Radio Shack ferric chloride "Etching Solution" for a few minutes/hours until all the surrounding copper has dissolved away. Take the board out and rinse it for a couple of minutes under cold running water. Be careful not to drip solution on any metal surfaces!!!!
Next, take some rubbing alcohol (higher alcohol content is better) and rinse off the sharpie ink with a paper towel. You can also buy some stuff called, "resist ink solvent", but rubbing alcohol works fine for me and its much cheaper. Finally, get some medium grain sandpaper and just rub off the layer of toner ink protecting the copper layer underneath. Simple!
(You'll also need a hobby drill like a Dremel or Black and Decker and a mini drill bit to make the holes. You could try to do it manually but the Dremel makes things much faster. You can drill all the holes in about 30 seconds.)
[size=small]Constructing the Amplifier:[/size]
The only problem I had with the amp was that from looking at KG's component placement diagram labeled, "Top White Print" (left), it appears that the top two transistors (circled in the diagram--1815 on the left, 1015 on the right, looking down at the board from above) are facing the wrong way.
That is to say, the flat side of the top right transistor (1015) should be facing the top of the board and the flat side of the top left transistor (1815) should be facing the bottom. The diagram has these two oriented the other way around. Other than that, populating the board should go pretty quickly.
Here is the corrected composite view/stuffing diagram of KG's Gilmore amplifier. Please note that in this picture, the transistors are oriented correctly.
Click here for a higher resolution image of the stuffing diagram
Also, if using KG's PCB design, you can download and then crop the image in Photoshop/Paintshop to get rid of the header containing the words, "Bottom Side Dynamic" and the footer with "3.800 x 3.300..." on it. This lets you resize the image to the specified 3.800" X 3.300" dimensions and print out at that resolution with no problems. You can fit two of the amp circuits side by side on a single large Radio Shack copper clad board.
One final construction note--I know that KG specified 1.6v LEDs from radio shack, but I could only find 1.7v LEDs so I used them and it seems to work okay. The measured dc output voltage is something outrageous like .0012 Vdc to 0.000 so the servo is something swell to have!
Here is some price information for MCM -- 800 543-4330
2SJ109 Toshiba FET $4.15 ea. (2 required)
2SK389 Toshiba FET $3.21 ea. (2 required)
2SA1015 Toshiba Trans. $0.18 ea. (12 required)
2SC1815 Toshiba Trans. $0.12 ea. (12 required)
I am also using the good old OPA134 as my servo opamp, not the OPA227.
For my fellow newbies out there, I want to say that I took a good long time to study the schematic before I understood what was going where, esp. in terms of the servo and where it connects back to the "dc adjust", etc. The great news about the Gilmore amp is that if you use KG's PCB design, you don't have to really worry about that...All you have to do is match the resistors and just solder the components where the diagram says they should go. Its really a great thing for KG to have made this great sounding amp/board and all the other design material available because its basically a cut and paste excercise for us non-techies(if you use his board!). Plus, it gives off a really cool glow--
Anyway, that's it for tonight and my apologies for my long-windedness. I'll actually get down to comparing the sound qualities tomorrow...