Re: tektronix 7S14 batteries and time base question

Bruce Griffiths

Hi Dennis

Some issues with the 7S14 sampler circuit schematic (1988 version)

1) The avalanche bias adjust circuitry shown in the schematic doesnt work. The 470 ohm resistor should be connected between the collector of the npn (Q20) transistor and HV.

2) The sampling pulse generator circuitry shown produces zero output.
The tapped shorted microstrip delay lines that shape the output from the avalanche transistor have been omitted from the schematic. These are obvious in the actual physical layout.

Any mismatch between the 2 series resistors will result in sampling pulse residue at the preamp input.

The sampling pulse width will be around 350ps or thereabouts (effective sampler risetime specified as 350ps).

Have started playing with an LTspice simulation (omitting blowby compensation for now).


On 27 December 2019 at 16:42 Dennis Tillman W7PF <> wrote:

Hi Bruce,
Thank you for your post on using clear epoxy to bond LEDs to each other. I chopped off the ends of a white source LED and a red receiving LED with a sharp razor and I glued them together with clear epoxy as he said to do. Preliminary results show an insignificant difference in the output voltage this pair generates. I will be doing more testing to determine this and other things I experiment with in the next few days. I will report my results to the group.

In the original 070-1410-00 7S14 manual I have (Dec, 1973) the resistors in series with the mercury cells (R1 and R2) are 200 ohms each. In the 1985 revision (Apr, 1985) they are 2,000 ohms. The change took place at serial number B030000 and above.

The mercury cells reverse bias the sampling diodes until a strobe pulse forward biases them. Strobe pulses are extremely short. So even under heavy usage the current drain from the cells will have a very small duty cycle. It would be helpful to know how long a strobe pulse lasts and what voltage they go to. Has anyone done this?

The low duty cycle means a very low overall current drain on the cell. But during each strobe the diodes are forward biased and it is important that they get all the forward current they need to conduct. The very low internal resistance of the mercury cell means it can provide any forward biased current the diode draws. The 2K series resistance of R1 (and R2) will limit the current surge to ~500uA which is fine for the cell but that is much more than the output LED of the mercury cell replacement can provide.

A small value low voltage capacitor could store the charge coming from the output LED and provide all the forward bias current needed by the sampling diodes when they are strobed. The voltage it stores is less than 2V. I would suggest it be a low leakage capacitor because of the minute current coming from the LED. Glass capacitors are a good choice since they are cheap and have virtually no leakage. As an example, a 0.033uF glass capacitor (like the ones shown in this link are no bigger than a 1/4W resistor. Unfortunately I could not find surface mount glass capacitors for some reason.

Has anyone mentioned what the sampling diode technology is? If so I missed it. Also, someone mentioned a few months ago the temperature coefficient of the sampling gates. I went looking for where it was mentioned but I couldn't find it. I would think it would depend on the sampling diode technology. If anyone can remind me what the Tc was and what technology the sampling diodes are (I would guess they are silicon Schottky diodes) I would appreciate it. It affect many things we should consider when designing our substitute mercury cell such as turn-on voltage, current drain, strobe pulse duration, temperature coefficient, etc.

At this point in time I am the beneficiary of many experiments and suggestions on replacements for the 7S14 mercury cells. Previous informal random experimentation seems to have recently shifted to detailed discussions of important design considerations. What would help me make a contribution of my own is 1) knowing the technology of the sampling diodes, and 2) documenting the strobe pulse's voltage, duration, and current drain on the battery. I'm hoping someone can help me with #1. I can do #2 myself.

I welcome comments since, as far as I know, some of my thinking hasn't been discussed yet.

Dennis Tillman W7PF

-----Original Message-----
From Bruce Griffiths
Sent: Monday, December 23, 2019 4:17 PM

LED encapsulation is clear epoxy. Clear epoxy is best suited to bonding LEDs end to end. Either mechanical preparation of the mating surfaces or priming is required. Done properly the roughened ends are thoroughly wet by the bonding epoxy and the joint is invisible.
The refractive index of cyanoacrylate adhesives is somewhat lower than that of epoxies.
For a Lambertian source like a LED, butt coupling is effective. Intervening optics doesn't improve the coupling. The dome lens on the LED doesn't collimate the light so that an intermediate lens is required to maximise coupling if this method is chosen.

For maximum coupling when butt coupling the separation of the 2 LED die should be comparable with the source diameter, However reflected light from the cupped lead frame of the LED may relax this requirement a little.
Optocouplers use butt coupling with a thin transparent insulating film between the emitter and the detector.
The peak response of a LED used as a photodetector occurs at a wavelength that differs slightly from its emission peak when used as an LED.
Most of this was covered in the HP optoelectronics Handbook around 1970-80.


On 24 December 2019 at 12:34 Dennis Tillman W7PF <> wrote:

Hi Chuck and John
I was surprised that several people ground the LED ends flat and glued them together. This seems counter intuitive to me for several reasons.

These are the reasons I think that grinding down the LED ends is not a good idea. I would appreciate it if you could explain the flaws in my thinking.
1) The polished surface of the LED lets the most light out. Wouldn’t a ground down (rough) surface scatter and block a portion of the emitted light.
2) The LED's dome shape focuses the light into a fairly narrow angle which increases the likelihood that the majority of the emitted light can be aimed right at the die of the LED that will convert the light to electricity.
3) Crazy Glue may appear clear to humans but what are its optical absorption characteristics? Does it absorb any of the wavelengths generated by the LED emitter?

On the other hand I think there are advantages to grinding the ends flat:
1) The ground end combination takes up a fraction of the volume of two unground LEDs.
2) Mating the two LEDs flat against each makes it easier to align them to each other.

It seems to me that the greatest conversion efficiency will come when you can place a bare emitting LED die on top of the die of the receiving LED. At that point every emitted photon can kick out an electron in the receiving PN junction.

IR light is another issue I'm confused about. I think I must have misunderstood but it sounded like some people think IR LEDs would make a good choice for emitters. Wouldn't just the opposite be true since a photon's energy, E, is proportional to its frequency, v, as in E = hv. Do IR LEDs emit more photons (greater brightness) and that is why they are a good choice? If so does the same thing apply for the receiving LED - which would have a high conversion efficiency resulting in the largest number of electrons being produced?

Dennis Tillman W7PF

-----Original Message-----
From: Chuck Harris
Sent: Monday, December 23, 2019 7:15 AM
Subject: Re: [TekScopes] tektronix 7S14 batteries and time base

Hi John,
That has been my experience as well. I did a long stint in a lab where we were doing IR spectroscopy, using lasers.
When I tried to make such a bias device, I ground both LED's ends flat, and welded them together with crazy glue. I figured that it would reduce reflections at the red I was using.
I couldn't get spit out of them... measured with a 200M input impedance meter... I guessed the older LED's just weren't bright enough.
Or, maybe the mechanism is not reciprocal?

John Griessen wrote:
On 12/22/19 11:30 PM, Chuck Harris wrote:
If I have been following correctly (always suspect), aren't we
using an LED illuminated by another LED to behave as a photo diode,
and produce the bias voltage for the switch?
One thing for sure from back when I worked with near IR LEDs and
laser diodes in a narrow beam system is that what absorbs IR or
reflects or not is not obvious from our visible light experience...
So, the efficiency could be because the incoming IR light "gets in"
instead of reflecting. They are both designed only to output, yet one is being used to receive...

Longer IR tends to go through more things that look black to us, and
probably go right through the plastic of LED lamps without much
refraction so angle and placement can be whatever.
Dennis Tillman W7PF
TekScopes Moderator

Dennis Tillman W7PF
TekScopes Moderator

Join to automatically receive all group messages.