16-05-2020, 04:35 PM
(This post was last modified: 16-05-2020, 04:55 PM by Mike Watterson.)
A mill is now preferable to etching PCBs at home. I did some double sided ones on 0.8mm and 1.6mm board in my last job. A cheap mill using a Dremmel type tool for the milling works. Use the mill system to drill location holes to flip the board and drill the pads, though mostly I was soldering to the tracks, SMT ICs, caps, transistors and resistors. Pre tin pads and use clean iron with even a 3mm chisel style tip that you'd use for ordinary parts. More accurate than most home etching. Used to use UV film, big UV box and decent etch tanks. The mill kits are not too expensive. Use a suitable bed fastened down on the main bed and mill it slowly all over to get a really level bed for thin flexible PCBs.
For more than I'd use China or Eastern Europe. Mostly I use matrix board and stripboard. No-one sees it inside the box! You can do the SMT on vero and even score down both sides of the holes to have 0.05" pitch tracks, though adaptors for 8 pin to 64 pin ICs of different pitches are cheap. Practice with the 70kHz IF FM SMT IC which are about 20c each.
Last matrix board was DC pre-amp to use a 100uA meter movement to replace the duff 30uA meter on a MkIV AVO valve tester. I was given two AVO transistor testers that used the same meter (though different ballast coil in the meter box). They were both O/C on the actual moving coil. I used one 1/2 of the LM358 with a pot fed by +/- reference to reduce the offset. You need a dual supply to do that, so the PSU was a clock radio transformer and two diodes to give + and - raw DC. Pair of TO92 5V regulators, 78L05 and 79L05 as I had them. The pot + /- supply resistors clamped with a pair of 1N4148, likely to track. Pot feeds unity gain buffer that is then the "virtual earth" for the - pin op-amp resistor. The + in pin then a 3.3K to 0V to simulate the original meter, but 1nF across it in case of RF. 100nF decouplers on every supply pin and both ends of the offset adjust pot. Then a gain fine adjust pot in series with opamp op load to drive the replacement movement. Which fitted the original case! The meter swing was different, so scanned both AVO scale and donor meter scale and used transparency on a layer in the GIMP to distort a copy of the AVO scale to match 1980s meter. Laser print and glue to reverse of meter scale plate. Assemble all and discover that even the original meter case zero adjust fits to the donor movement slot!
Mk IV now working and you'd need to see it side by side to realise the meter is different. The end of the needle is just short of the top scale, which is electrode leakage or heater resistance, that's not a problem. The needle swing is a larger angle, hence the new scale needed.
Yes, I know there are real Instrumentation op-amps that don't need the offset adjustment, and ones with the offset pins. But I didn't have any. In practice the 2nd half of the dual opamp was otherwise doing nothing, so the only extra parts were a pot (needed on some opamps anyway), two diodes and two resistors.
Some makers application notes do show this idea of a voltage follower feeding a correction offset voltage. They all neglect to mention the offset pot needs low voltage regulated + & - supplies!
I tested supply sensitivity even with the ends of the pot diode clamped and saw that the 78L05 and 79L05 would do for main supplies, but only to power the opamp.
I also built a 30 LED bargraph at about 0.075V per step. The recommended minimum is about 0.065 for the IC. As two adjoining LEDs light about 1/2 brightness 1/2 way between steps the resolution is actually 1/60th full scale. If I used that as a meter for the AVO MkIV valve tester it would have been good enough, but needed a separate 5V regulator. I only did it on the plug in breadboard with green LEDs. I added a blue and red at each end driven by a dual comparator using the 1st IC reference at 1/20th and the main reference with a bit off. Only the 3rd IC's reference was used for the daisy chain of resistors for all three ICs Vhi and Vlo, except the 1st Vlo goes to 0V.
Because a real meter you can see if the voltage is zero or negative, or if it's past full scale. Hence the dual comparator for 1/2 of 1/30th / negative and for over FSD by about 0.05V.
Then I found the 100uA meter. I put a series resistance so it's FSD in volts was about the same as the LED bargraph. The Bagraph ICs and either skinny rectangle LEDs or SMD LEDs would have easily gone on veroboard and fitted in place of the meter movement and scale. The LEDs, SMD or skinny leaded ones, shine through paper well enough for a laser printed scale.
For more than I'd use China or Eastern Europe. Mostly I use matrix board and stripboard. No-one sees it inside the box! You can do the SMT on vero and even score down both sides of the holes to have 0.05" pitch tracks, though adaptors for 8 pin to 64 pin ICs of different pitches are cheap. Practice with the 70kHz IF FM SMT IC which are about 20c each.
Last matrix board was DC pre-amp to use a 100uA meter movement to replace the duff 30uA meter on a MkIV AVO valve tester. I was given two AVO transistor testers that used the same meter (though different ballast coil in the meter box). They were both O/C on the actual moving coil. I used one 1/2 of the LM358 with a pot fed by +/- reference to reduce the offset. You need a dual supply to do that, so the PSU was a clock radio transformer and two diodes to give + and - raw DC. Pair of TO92 5V regulators, 78L05 and 79L05 as I had them. The pot + /- supply resistors clamped with a pair of 1N4148, likely to track. Pot feeds unity gain buffer that is then the "virtual earth" for the - pin op-amp resistor. The + in pin then a 3.3K to 0V to simulate the original meter, but 1nF across it in case of RF. 100nF decouplers on every supply pin and both ends of the offset adjust pot. Then a gain fine adjust pot in series with opamp op load to drive the replacement movement. Which fitted the original case! The meter swing was different, so scanned both AVO scale and donor meter scale and used transparency on a layer in the GIMP to distort a copy of the AVO scale to match 1980s meter. Laser print and glue to reverse of meter scale plate. Assemble all and discover that even the original meter case zero adjust fits to the donor movement slot!
Mk IV now working and you'd need to see it side by side to realise the meter is different. The end of the needle is just short of the top scale, which is electrode leakage or heater resistance, that's not a problem. The needle swing is a larger angle, hence the new scale needed.
Yes, I know there are real Instrumentation op-amps that don't need the offset adjustment, and ones with the offset pins. But I didn't have any. In practice the 2nd half of the dual opamp was otherwise doing nothing, so the only extra parts were a pot (needed on some opamps anyway), two diodes and two resistors.
Some makers application notes do show this idea of a voltage follower feeding a correction offset voltage. They all neglect to mention the offset pot needs low voltage regulated + & - supplies!
I tested supply sensitivity even with the ends of the pot diode clamped and saw that the 78L05 and 79L05 would do for main supplies, but only to power the opamp.
I also built a 30 LED bargraph at about 0.075V per step. The recommended minimum is about 0.065 for the IC. As two adjoining LEDs light about 1/2 brightness 1/2 way between steps the resolution is actually 1/60th full scale. If I used that as a meter for the AVO MkIV valve tester it would have been good enough, but needed a separate 5V regulator. I only did it on the plug in breadboard with green LEDs. I added a blue and red at each end driven by a dual comparator using the 1st IC reference at 1/20th and the main reference with a bit off. Only the 3rd IC's reference was used for the daisy chain of resistors for all three ICs Vhi and Vlo, except the 1st Vlo goes to 0V.
Because a real meter you can see if the voltage is zero or negative, or if it's past full scale. Hence the dual comparator for 1/2 of 1/30th / negative and for over FSD by about 0.05V.
Then I found the 100uA meter. I put a series resistance so it's FSD in volts was about the same as the LED bargraph. The Bagraph ICs and either skinny rectangle LEDs or SMD LEDs would have easily gone on veroboard and fitted in place of the meter movement and scale. The LEDs, SMD or skinny leaded ones, shine through paper well enough for a laser printed scale.
