pchristy

Way back in 1968, I bought my first set of proportional radio - a Bonner Digimite-8, which I bought 2nd hand from a fellow club member. It served me well for many years before being retired from active service, but was occasionally pulled out for club nights to show "how we were"!

Recently one of the servos has started to act up, and I'm having some problems figuring out how its supposed to work! I have a circuit diagram for it and can figure out most of it, but some of it is has me completely baffled! I'm hoping Jaymen or someone else with experience of these sets can help me out!

Here's the circuit I have:

The problem is that the servo is constantly driving in one direction, regardless of the position of the output arm. I thought I had found the problem when I discovered that one of the 5uF capacitors in what appears to be the pulse stretcher circuit had gone leaky - electronically, not physically! I have some old scrap servos I acquired a while back for spares, and one of these yielded a replacement that tested OK, but it hasn't cure the fault.

The problem I have is that I can't work out how Q6 and Q7 are supposed to work! Q6 appears to be a half-shot, whose period is determined by the feedback pot and the 50K trimmer - pretty conventional so far. And indeed a scope shows that the collector of Q6 does indeed have a low amplitude pulse (amplitude limited by the base of Q7) whose width is determined solely by the position of the feedback pot. I would have expected Q7's collector to have an inverted version of this pulse, to be fed to the "summing junction" formed by the primary of the transformer. However, the output of Q7 is, in fact, an inverted copy of the input pulse from the receiver, whose width tracks the input pulse - not the reference pulse - and whose *amplitude* is governed by the feedback pot.

Now I can see how one or the other of these scenarios would work, but not both together!

If the pulse width is meant to vary, then when there is no error, both ends of the primary winding of the transformer will go up and down together producing no output. If one pulse is longer than the other, then an output pulse will appear on the secondary, whose polarity depends on which pulse is longer, and will therefore turn on Q2 or Q8 , in turn switching on the rest of the output chain.

Similarly, if the reference pulse simply tracks the input pulse, but with a varying amplitude, a similar scenario ensues. But why is the output of Q7 following the input and not the reference pulse? What am I missing here?

Comparing all this to a working servo doesn't really help much, as it produces an identical signal. The only place I can find a difference between the working and non-working one is on the pulse stretcher capacitors. On a working one, there is nothing when there is no error. When there is an error, one capacitor or the other has a sawtooth waveform on it - as I would expect!

On the faulty servo, one capacitor has a permanent sawtooth regardless of input or reference, whilst the other has negative going spikes.

I'm not helped by the fact that I have no component placement diagrams, and I'm having to backtrace some of the circuitry. Not easy when some of it is smothered in potting compound! Also, the scrap servos I got appear to have a later circuit board, although the circuitry looks to be the same. The later boards have the 50K preset re-positioned to be at right-angles to the wiper board, rather than parallel to and against it. Presumably this was done to make setting up easier.

I could probably fix this if I knew how this circuit was supposed to work, but from my examinations it seems contradictory and illogical. Can someone explain it to me please?

--
Pete Christy

tommav

My Feedback: (3)

Pete,
I had a similar problem with one of my Digimite 8 servos and went through the same frustration because the problem was a component you normally wouldn't suspect. Turned out to be a silicon diode on the wiper board. It still tested as a diode (somewhat) but the the low or the high resistance side was "way off" (sorry, I don't remember exactly as this was a few years ago). Might be best to just replace it. Hope this helps and good luck.
Tom Mavracic

pchristy

Tom, Thanks for the tip! I'm away from home at the moment,but should be back later on Monday. That diode looks as if it is the one for the fail-safe, but I'll know for certain when I get the servo back in my hands. I'll keep you informed!

Cheers,

Pete

ukengineman

Hi Pete, a few years ago I restored an F&M Digital 5 which used Bonner Digimite servos. I had problems with all the servos and there was no single fault pattern. I also found that the circuit you posted (I think it came from me) was obviously wrong around the reference pulse generator. I looked at the notes that I scribbled on the circuit at the time and see that I noted that D2 was shown wrongly but unfortunately I did not record what the true position of the diode was. I do recall that I had a couple of the 2N2712 that were faulty and one case of a failed GC4066. There may well be other differences between the actual servo and this circuit. One thing I learned very quickly when disassembling Digimite servos is to do it on a tray to catch any ball bearings that try to escape! It would be interesting to create a Spice simulation of the circuit, maybe I will do it one day.
Alan

tommav

My Feedback: (3)

Hi Pete,
Yes, the diode is associated with the Fail Safe feature. As I recall (and again it's been a few years now), my servo traveled to an extreme end (or nearly so) with the transmitter on, but returned to the neutral position with the Tx turned off. It was as though the Fail Safe still worked but the neutral position (with signal) shifted dramatically. I also refurbished a Digimite 4 channel back around the same time so there's a chance I may be telling you about a 4 channel servo, but I'm 90% or better certain the problem was with an 8 chan servo.

The circuit you have came from a 1969 publication (first printing was 1968) called Radio Control Manual by Edward L. Safford, Jr. I added the "Bonner Digimite 8 servo" label as the publication has it incorrectly shown as "Digimite receiver and decoder" in Figure 1203. Not Mr. Safford's error - was the publisher's. The other Digimite 8 schematics on the website also came from this source.

Tom

pchristy

Hi Tom and Alan,

Thanks for the support, guys! I've been away all week-end visiting my son 260 miles away. We had to go in my wife's "shopping trolley" because my car decided to let me down for the first time ever just before we left. I think its going to take me a week to straighten my back....!

It wasn't the diode Tom, though I can see how that might cause an issue if it was leaky. Actually, the whole fail-safe circuit isn't working at the moment - probably the big electrolytic associated with it, but I've been concentrating on the servos first before tackling the receiver.

Seems like everyone back around that time was using Bonner servos. I know F&M did, and the scrap ones I've got have little "RCS" badges stuck on them! RCS were a British manufacturer and came out with an F&M clone - the RCS DigiFive - which also used Bonner servos. The RCS ones are all completely dead, and look as if they've been in the wars a bit, but they do provide a useful source of spares! And yes, I'm very well aware of the ball-bearing issue! Back in the day, when I was flying it regularly, I used to have to strip and clean the servos at least once each season - usually twice! I got quite adept at replacing the ball bearings and used to have a special pewter tray I used for stripping them. Funny, though, just shy of fifty years on I don't seem to be able to re-assemble them anything like as quickly as I used to! I wonder why that could be......?

The fault is definitely in the vicinity of the pulse stretchers. As far as I can tell, the feedback reference generator is working correctly - even if I don't understand how its *meant* to work! It is, at least, the same as a functioning servo! That means its either one of the initial switching transistors - Q2/9 - or maybe one of the subsequent transistors has gone leaky Q3/9.

I'm a bit busy with other things - like getting the car fixed and the wife's birthday - at the moment, but I'll be getting back into fault-finding mode soon. Maybe at the week-end - the weather forecast is awful - again!!!

--
Pete

HighPlains

My Feedback: (1)

I always assumed that the early F&M used just the Bonner mechanics and their own electronics. I have a slightly later Digital 5 with the Kraft KPS-7 mechanics. But that was in the early 7 cell era when there was lots of sharing of ideas and few customers. On the 7 cell F&M sets there were two different battery set-ups, one with +/- 2.4V to the servo motors and another with +/-3.6V to the motors.

On your system I would expect to see a brief spike with the leading edge of the control pulse at the transformer, then the longer error pulse as it charges a capacitor when the error feedback pulse is formed. I had heard of this amplifier described as an analog servo amplifier, and the drive circuit is to a degree pretty close to an analog servo, the charged caps have to decay somewhat to a point where the transistors in the drive shut off.

It is certainly not a PID controller with a high frequency error integrator giving rapid fire pulses to the final drive circuit.

ukengineman

The only parts of my F&M Digital Five Bonner Digimite servos that appeared to differ from the schematic in any significant way were the reference pulse generator and the input buffer. This may well be due to the difference between the Bonner double short pulse versus the conventional single pulse of F&M

jaymen

When the incoming pulse fires the one shot comprised of Q6 & 7, Q7 collector goes low low for the duration of the time constant set up by the pot wiper. A positive going pulse appears at one side of the transformer, and the negative going pulse from the one shot is on the other end of the transformer primary winding. If both pulses are of the same duration, they cancel each other out. If the negative, or positive pulse is of greater duration, it creates a negative or positive pulse at the output of the transformer secondary. The pulse has to be long enough to charge the 5uF (4.7uF) caps to trigger the transistors onto conduction. If the failsafe bias diodes are leaky, the amp will drive in one direction all the time. You could also have a bad transistor, or a leaky 5uF cap that would do the same thing, and a bad wiper or feedback pot would also cause the same problem.

pchristy

Thanks for all the info, folks, and to Jaymen for his explanation of the reference generator / comparator. I'm not sure that's the whole story, because there are some other very strange things happening around that area, but it did convince me to have another look at the reference generator.

As an aside, the fail-safe suddenly started working again! Maybe the electrolytic reformed, now its been switched on for a while again! Whatever, that convinced me that the fault preceded the failsafe circuitry in the servo, because the servo failsafe'd perfectly in the absence of a signal. Therefore the drive circuits must be OK.

Opening up another servo that works - something I hesitate to do because of all those ball bearings (!) - I noticed that the reference pulse in the broken servo is much shorter than in the working one. This is pointing me towards the reference generator. My next step will be to have a look at D1 and D2 (possibly pulling the volts on the C3 charging circuit down due to leakage) - or maybe C3 itself.

Haven't had much time to work on it this week due to higher priority domestic issues - and also preparing for the single-channel and retro meet in Pontefract in a week's time! It would be nice to get the Digimite working for that!

As another question, the output transistors are GC4066s. I know these were specially made for Bonner - basically a standard Texas transistor with an added heat sink. Presumably the heatsink is needed because the transistors are working in an "analog" mode, ie: not fully on or off. I don't know what the original transistors that these are based on were. I know they are PNP, and suspect that they may be a germanium type. Can anyone confirm this? I ask because I have some small silicon power PNP types that are rated at 1W dissipation that would probably make good replacements if and when the time comes, but if the originals are germanium, that would presumably upset the working points.

Onwards and upwards......!

--
Pete

ukengineman

Hi Pete, one of my F&M Digital 5 Bonner Digimites had one of the GC4066 motor driver transistors dead. A few tests showed that they are germanium devices but I had 100% success replacing the faulty device with a 2N5322 silicon transistor (1W rated). In my case it was Q5 that had died and you will see that the way it is used in the circuit means the higher Vbe is not really a problem, I'm not sure a silicon device could be dropped straight in for Q11.

Good luck, Alan

jaymen

You can use a 2N4403, or a ZTX 750, or ZTX 751. Replace them as pairs if you do, so the amp is balanced.
You should replace both of the 5uF using 4.7uF caps of the same brand, so they are matched. You can disconnect the fail safe diodes if you think they are causing a problem.

If you look closely at the one-shot (reference pulse generator) you will see how the transformer primary works as a summing junction. On the secondary side, you get the difference in pulse width, and polarity, to drive either direction.

Make sure you don't have an end failsafe servo, which would be throttle.

I would also replace all the electrolytic caps, especially the big one in the failsafe circuit, as when one goes, they all are suspect. Many times, I have fixed radios with weird problems by just replacing the electrolytics, it's amazing.

We used to take apart the Digimite servos in shoe box to capture the balls. Use a dab of silicon grease to keep them in place when assembling.

pchristy

Thanks again, folks. Alan: Yes, I had a feeling they were germanium types. I don't think there were many power silicon transistors around just then! Jaymen: I tried replacing one output transistor (Q5) with a ZTX 753, and that worked fine, but I suspect that, as Alan says, Q11 might be a different matter due to the higher volts drop of a silicon emitter follower. You only have 2.4 volts to work with in the first place, so an extra 0.4 volts drop is going to make a significant difference, I would have thought.

Moving back to the reference generator, yes, I understand how the transformer works as a summing junction (see my original post). What puzzles me is that the feedback pot seems to vary the *amplitude* of the reference pulse rather than its width. Maybe its partly because the half-shot isn't working correctly, but at the moment, the output of Q7 is an inverted version of the input pulse (as Q1), but whose *amplitude* varies with the feedback pot. Maybe when I get the reference half-shot working correctly it will all become clear. I'm just about to have another crack at it......

And yes, I know about electrolytics! When I was a trainee, I was attached to a very wise and old-fashioned maintenance man who taught me to "1) Check the power supplies, then 2) Check the electrolytics. That will cover about 90% of the faults you will come across!" He hasn't been far wrong to date!

--
Pete

pchristy

FOUND IT!

Would you believe it? It was C3 - the .047 reference timing capacitor! It looks as if one leg had become detached inside the capacitor, because when I de-soldered it, one leg fell out on its own! Only the potting compound was holding it in place! Ye Gods!

Anyway, I've learned a lot more about the workings of the Digimite servo amp - more than I probably wanted to know!

BTW, regarding the "double pulse" system, I'm still puzzled as to why they adopted this approach. The servo only uses one of the "double" pulses - the other is just wasted. Maybe it was to keep the frame rate constant to help with the pulse stretching circuitry, which is certainly a bit unconventional.

Back in the day when I was still using this system, I made my own receiver to go with it. The receiver deck was a RCM Classic, but with a home made 16 channel decoder on the end, built using them new fangled CMOS chips! It worked extremely well, and a useful function was that the even numbered pulses moved in the opposite direction to the odd numbered ones. To reverse a servo, you just moved it forward (or back) one output. The very slow frame rate made the servos noticeably slower than on a conventional system, but hey! It worked. And still does! I used to fly a Kavan JetRanger and a Schluter Gazelle with that receiver and the good old Digimite tranny. For its day, it had phenomenal power output!

I recall a fellow club member buying a Digimite-4, because the 8 had been seen to be so reliable, but had serious range issues with it. Our club "boffin", the original owner of my Digimite, had a look at it, and found virtually no output from the Tx! A quick tweak with a tuning wand got it up to where it should have been, but the owner had lost confidence in it and sold it to the boffin for a song! The boffin continued to fly it for a couple of years, before passing it on to another club member who flew it reliably for many years thereafter. Happy days!

Anyway, thanks everyone for the help and suggestions - not to mention the moral support!

--
Pete

HighPlains

My Feedback: (1)

Good show. I have heard the term Boffin before, but never looked it up till now. This in spite of working with a couple of blokes at a east coast tech company for a couple years.

ukengineman

Well done Pete, we've got to keep these old systems working! My last project was to get an Orbit 4 channel analogue system working again but that is another story.

Alan

jaymen

The original Digimite 8 double pulse encoding scheme had to do with the fact that they were using only the leading edge of the pulses for triggering and timing. This was before TTL, where you have an NPN, and a PNP transistor to turn on and off the pulse. These early designs used RTL (resistor-transistor logic) So because the turn on time of the transistors was much faster than the turn off time, to keep the slope of the trailing edge of the pulses from causing neutral drift, they did not use the trailing edge, and instead, used the leading edge of a second pulse to trigger each 1.7 msec flip flop on and off.

Later designs used a clever 250usec one shot, that like the reference pulse generator you just fixed, had two transistors and created a clean falling edge that could be used for reliable no-drift timing of the pulses.

pchristy

Jaymen: Yes, that makes sense! I sometimes forget that this system was pioneering the outfits we have today! 20/20 hindsight is a wonderful thing!

My last project was getting my old Grundig "reed" (actually electronic tuned filters) system up and running after a prolonged period of idleness. Most of the issues centered around a wealth of contacts that needed cleaning, although the progressive throttle servo needed a fair amount of freeing off! That one is now working extremely well, and if I could find a club that would permit 27MHz, I'd be quite happy to fly it!

Not so sure about flying the original Bonner receiver and servos. The seven-wire connectors are not as solid as they once were, and sometimes need a bit of wiggling to coax things into life. I'd be quite happy to use the tranny with my home-built airborne pack though......

Mi Digimite:




And my Grundig "reed" set:


Here's my current system: An Ace MicroPro single-stick, converted to 2.4 GHz with a FrSky "hack" module:



I'm currently flying a 1963 aerobatic design called a "KingPin" with it, using a pre-Blackhead Webra .61 that I bought around the same time as the Digimite. The "KingPin" was designed by Dave Platt, before he became a scale guru and emigrated to the USA. Here's some video of it on its maiden flight:

https://www.youtube.com/watch?v=ibqjtqprJZk Happy days!

--
Pete

jaymen

Ah yes, Dave Splatt's Kingpin, a real classic. I recently repaired a 6 meter Ace Silver series receiver for Dave, he flies a MP8K like yours.

We have taken several Digimite 8 radios and replaced the encoder with an Airtronics, or a Hitec Focus. You get servo reversing, D/R, travel volume, and your choice of AM/FM/ or SS for a module. The later "RS" Digimites use convention PPM, so all you need to do is go buy an HPI, Futaba, or Traxxas 27MHz AM receiver and some servos and you're ready to go fly.

pchristy

Jaymen: There's quite a movement here in the UK for updating old radios with modern internals - I think I've seen you posting on our single-channel site? http://singlechannellersreunited.co.uk/ However, the Digimite is quite a rare beast these days, and I'd like to keep it as original as possible to show people where we came from! The home-made receiver works absolutely fine with it, so no need for any modifications to the transmitter.

I remember seeing Dave Platt flying at our Nationals, back in the 60s, when he had become something of a legend in the scale competitions. He was one of the first to apply the "weathered" look to his aircraft, rather than building them looking factory fresh. He was also one of the few scale pilots back then who could really fly as well as build!

If you speak to him soon, let him know he is still fondly remembered on this side of the pond, and that his designs are still being built and flown!

Cheers,

--
Pete

John Schall

After all the time that has passed, it doesn't look like you need further help, but just in case, I do have a working schematic for a digimite 8 servo. I have the same problem with Safford's book. I suspect that the problem with typos extends to the receiver as well. I found errors in TX scam. I reversed engineered a working servo to get it. I don't have a scanner so the scem will have to wait a few days. A few things I found out about the servo that bear mentioning. I don't have a working original Bonner 8 ex, but I strongly suspect you don't see an oscilloscope signal at the servo pigtail unless a servo is plugged in. The voltage that causes the diodes to conduct comes from the servo. The first stage of the servo scam in Safford's book is right, so the three 27k resistors form a voltage divider. I had 2 servos that worked, 5 that didn't, so it was worth the work. Now all 7 work. The large .047uf cap should not be replaced with a disc, if you do, the sv will work, but it will wander and be unstable. There are many transistors that will work, in one sv I had to replace all of them. I used 2n3904 for NPN (9) and B772 or was it D882 for the PNP. To test the d.c. part of the circuit, with no sign coming in, short out one of the 5uf caps. Servo should go to one end. Other cap will cauze it to go to other end. If not do, then problem is between motor and transformer. More later.

John Schall

I had to reverse engineer a working servo to get a good scematic. Safford's scem isn't too screwed up, just a few fixes needed.First, disconnect collector of Q7 from rest of ckt and reconnect it on the other end of D2. Then add a 2.2k resistor in series with D1, before or after, it doesnt matter which. Not all servos need these next two things, but I see them a lot. A 27k feedback resistor from the green wire on the motor to the pot wiper and 1uf electrolytic cap bypassing the wiper to ground.

scem, bon digi 8 sv