Neutron optoisolators status report

[armillary]

I recently built a Neutron using the GGG PCB and the build documents on the GGG website. The build is straightforward enough, and as others have noted there's an error in the wiring for the gain pot, and the 7660 charge pump has a whine, but those are well documented. If there's one subject where there's lots of frustration with Mutron-III clone builds, it's on the subject of optoisolators. So I'm providing a status report on that, and welcome your comments.             

When building the Neutron, socketing all the ICs and using a pot for Rx (the current limiting resistor), is a must in order to find the best sound (or any sound at all). For Rx, I used a 100K ohm pot in series with a 1K ohm limiting resistor. The Neutron build guide mentions an Rx value of less than 4.7K ohms. That might be true if you're using LEDs from 1972. More on that in a minute. 

I made my own LED/LDR optoisolators, each using one 5mm LED and one LDR stuffed in opposite ends of a short piece of 5mm diameter black shrink tube. The rounded end of the LED is face to face with the sensor of the LDR. A 5mm disk of shrink tube is used at each end to keep the light out. (Use a hole punch to make the disks from a larger piece of shrink tube. Use a 1mm drill bit to drill holes in the disks to match the leads of the LED and LDR.) Slide the disks over the leads, insert the LED and LDR in the 5mm shrink tube and line up the leads so they're all on the same plane (like a capital H). Then use the heat from a soldering iron to shrink the ends of the shrink tube so it's light-proof and everything is held in place. They're like home-made NSL-32s. They look like crickets. A matched pair of these is soldered into an unattached 14 pin IC socket, illustrated as Option 3 in the Neutron build guide. The socket makes them modular, so you can plug that socket into the empty 14 pin socket you installed on your GGG PCB. I've made 6 of these modules so far, and I'll probably make a few more.     
 
For the LDR, I only used the Waitrony KE-10720. Others have used it, it's cheap and it's the only LDR Tayda sells. The KE-10720 LDR has a wide range, from a dark resistance of several megaohms to a light resistance of less than 1000 Ohms. It's very sensitive, so it only takes a small amount of light to drive it to the lower limit of its resistive range. It's designed to work with a light wavelength of 650nm, which is in the red light range.

For LEDs, I chose the standard 5mm size referred to as the T-1 3/4 package. I only used diffused red LEDs in my testing. The main variable in choosing the LEDs was brightness.

The Mutron-III was first made in 1972. Most of today's LEDs are MUCH brighter than the LEDs of 1972. LED luminous intensity is measured in Millicandela (mcd). Modern diffused red LEDs are rated at 8000 - 10000 mcd. Water clear red LEDs are 12000 mcd. Even at very low current, from the instant they reach their forward voltage, most modern LEDs produce a huge amount of light.

LEDs in 1972 were likely rated from just a few mcd to several hundred mcd, a tiny fraction of the intensity of most modern LEDs. I tested several brands of T-1 3/4 style red LEDs, and I found that the lower the mcd rating, the better results I got in the sound and playing dynamics when testing in the Neutron. The module I made with 10000 mcd LEDs barely quacked with Drive set to down and the CLR set to 80K ohms. The module with 200~300 mcd LEDs quacked much better, and the CLR could be set to 40K ohms. The module with 1.7 mcd works the best so far, not just for Drive down, but for Drive up as well. With the CLR at 22K ohms. And it even works well with bass guitar. So I think we're getting close. Less is more.

What's the problem with the super bright modern LEDs? I did some testing, and I think the problem is that even at their lowest current, they're so bright they drive the KE-10720 to below 10K resistance. And any more current and the KE-10720 is below 1k ohms. So the operating range is only about 10K. At high current, the low mcd LEDs will also drive the KE-10720 to below 1k, but when the low mcd LEDs are dim they are very dim, the KE-10720 will show a much higher resistance, and the operating range will be much larger. Low mcd LEDs seem to be a better match for the KE-10720, so you get better envelopes.       

You can find low mcd T-1 3/4 style red LEDs, you just have to look at the specs or packaging to find them. Some are NOS, but there are new ones too. The 1.7 mcd LEDs I just tried in the last few days are NOS NTE3020 models I found on ebay. The prices on ebay are all over the place, but don't spend more than 50 cents or a dollar per LED for these. On Mouser you can sort LEDs by package, color and luminous intensity, down to 1 mcd. The spec sheet for the WP7113HD LED that Small Bear sells (for 25 cents each) lists them as 5 mcd. I plan on trying some of these soon.         

[armillary]

While I'm waiting for more LEDs to arrive (I ordered some Vishay TLLR5400 to try), I have a question about using one LED per LDR, versus one LED for both LDRs. I assume the Mutron-III used a single LED, based on the original 0805 having one pair of LED leads. (Note there could be two LEDs inside the 0805, but we'd have to cut one open to find out.) The Neutron has the option of using two LEDs, and that's what I've been testing. But I've noticed that the PCB connects  the two LEDs in serial. That makes the forward voltage (Fv) the sum of their individual Fv values. I've tested it on a breadboard, they light at 1.7V individually, but in series it takes 3.4V to get them to light. Or in parallel they light up at 1.7V. The Neutron works fine, so maybe it's a non-issue, but I'm thinking about the voltage range of the LED driver circuit (U3a) and wondering if losing the range from 1.7V to 3.4V could have any shortcomings. I'm thinking about a PCB mod to wire the LEDs in parallel to see. Is this something that's been covered before?         

[Mark Hammer]

My understanding is that it used a Hamamatsu unit.  I haven't installed it yet, but I bought mine from Small Bear a while back.

Something I've done in past to use two LDRs with one LED is to use one of those rectangular LEDs.  The flat sides allow for the LDRs to be placed flush against the sides of the LED, with a bit of superglue to hold them in place while I place heat shrink over the assembly.

Note that "regular" cylindrical LEDs come with a lot of excess plastic that isn't required for the LED to function properly.  I grind/file the round ends of LEDs down until there is only enough plastic left to safely protect the anode/cathode assembly.  The last step should be sanding with a medium that provides a smooth and relatively consistent surface on the LED end.  That helps to diffuse the light produced.  It also provides a good surface for the glued LDR to adhere to.

[PRR]

There was an opto-part with a split LDR on one LED. That may be the original.

To a first approximation, IC3a will slew to 1.7V, 3.4V, whatever it takes to get some response. Yes, in a 9V world the opamp may run out of swing first.

Since LEDs have got SO much more efficient, you can put them off axis (off the bright spot), or increase Rx to decrease current. 4.7K may _not_ be too large today.