Max Frequency of SM 118A counter?

Anyone know the maximum frequency of a Heathkit SM-118A? I have been all over Google with little luck. One guy said it would go to 1 GHz maybe 2 GHz, I don't think so. Mikek

Reply to
amdx
Loading thread data ...

As I vaguely recall, none of the early Heath LED counters went much above about 500 MHz.

I couldn't find a free schematic or manual. Open it up, look at the prescaler chip, grab the Heath part number, and find the real chip number from a parts cross reference. Then, lookup the specs on the chip, which sets the upper frequency limit. For example, if the counter section works to 50MHz, and the prescaler is a divide by 10 type, then your upper limit is 500MHz.

Totally useless YouTube video of the SM-118A in action:

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Do you have a general idea on vintage?

Their first frequency counter, out in 1971, went to 15MHz, 20 if you were lucky. But external prescalers soon followed, but they were generally divide by ten, so the upper limit was determined by how fast the prescaler could count, but also the upper limit of the counter itself. Actually, in time Heathkit sold a matching prescaler, a box the same size as the counter.

Likely the second generation of Heathkit counter had a prescaler, I can't remember. But timing is everything, if it came out shortly after the first one, it probably had limited range, not even 1GHz. COming out later, you could put a 74S196 counter as your first counter and get up to

50MHz, which then meant a divide by ten prescaler would give coverage to about 500MHz. It took some time before the cheap frequency counters got up to where they read fairly high without a prescaler, and by then the prescalers had better high frequency ability.

I reemmber the model number, but I'm blank about the period or specs.

Michael

Reply to
Michael Black

amdx schrieb:

It may be helpful to ask

here:

HTH

Reinhard

Reply to
Reinhard Zwirner

I'm waiting for the administrator to decide I'm Ok, and let me in. Mikek

Reply to
amdx

This is kinda funny, hardly a mention of the SM-118A found on Google, but some how I end up with two of them. Both for sale btw, going to the Orlando Hamfest in Feb. Mikek

Reply to
amdx

"Michael Black"

** The Fairchild " 95H90" was one of the early ones, dividing by 10 or 11.

Came out in the mid 1970s.

Upper limit was about 250MHz.

Ones that went to over 1GHz were much later and used the "Josephson Ring" idea - so they divided by numbers like 64 or 128.

... Phil

Reply to
Phil Allison

The 11C90 was one of the first dual modulus prescalers. I used it in a VHF and an HF marine radio synthesizer. It worked great. I can't say the same for the 11C44 phase/freq detector which provided a challenge minimizing the dead band and reducing jitter. The problems would have been tolerable had the chips from the various Fairchild factories performed identically. Unfortunately, they did not. We resorted to tweaks specific to the 11C44 country of origin. That kludge held things together long enough for us to switch to the Motorola MC12013 (10/11) prescaler, MC12014 control chip. There was also an MC10138 (2/5/10) doing something useful in the synthesizer. The change would have happened sooner but the company was a spinoff from Fairchild and such loyalties tend to die slowly.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Ooops, memory fault. I just excavated the schematic of the Intech M3600 HF SSB radio. The synthesizer has the aforementioned Motorola ECL prescalers, but the synthesizer phase detectors are still the Fairchild 11C44. I could swear that I killed that monster, but I guess not.

You can see the problem in Fig 1 on Pg 229: The middle of the non-linear region is essentially flat. That means there's no change in phase detector output voltage for changes in input phase. The result is that the oscillator frequency just bangs around between the phase detector dead zone limits producing rumbling noises in the tx and rx audio. There were several tricks for reducing the problem, but all had undesirable side effects.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

** Yep.

** Not so useful at 1 or more.

To interface with "ordinary" logic families, division by 128 or 256 is needed.

Like the Philips SAB6456 used in millions of CTV sets.

formatting link

.... Phil

Reply to
Phil Allison

"Phil Allison"

Reply to
Phil Allison

** Not so.

Philips described it as a Johnson ring in the app notes.

FYI:

With no input, SAB6456s oscillate at 1.35 GHz.

Rest of you smug, context shifting bullshit snipped.

Bye....

... Phil

Reply to
Phil Allison

Can't remember. It was nearly 50 years ago.

Late 1960s/early 1970s. The ring counter was never a problem. All TO-18. Never needed to touch that. The bad plastic was the succeeding dividers,

20MHz downwards,each on its own lille PCB. Early plastic stuff that were notorious for bad passivation. We used BSX20 simply because it was on inventory in bucket loads. After the first few fell over, we just changed out the lot.

Don't get hung up on ft. Device capacitances, tr, tf, tstorage, are where it's won and lost. Bode plots aren't the whole story. There are plenty of

900MHz ft devices that suck as switches.
--
"Design is the reverse of analysis" 
                   (R.D. Middlebrook)
Reply to
Fred Abse

Yeah, 4046es still have that problem. Usually a 1 M resistor from PD2 output to ground fixes it, because all you have to do is pull the quiescent point a few nanoseconds from the null.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Mixing is not that horrible. That's the way my HP5246L and HP5248M counters work. The plugin has an input bandpass filter and a mixer that down converts the input to 0 - 100 MHz. The local oscillator generates a comb line with 100 MHz spacing. Basically, the counter repeats the count every 100 MHz. There's also a signal level meter to help tune the preselector and set the input level. The bad news is that I can't sync either the counter clock or the local oscillator to an external frequency reference.

It works quite nicely as a cheap microwave frequency counter as I have

8 of the 12 available plugins. Also, Nixie tubes are cool.
--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

To the best of my limited knowledge, it's a problem with all digital phase detectors. Anything that goes tri-state (i.e. open circuit output) when locked is going to have a dead band. I found that out the hard way when I tried to replace the 11C44 with a D flip-flop equivalent, and had the same problem.

Using a large value resistor was one of the first things we tried. The problem was that any offset in the loop amplifier would dramatically increase the reference oscillator feedthrough.

The 1M resistor is also a problem in marine radios. One just does not use large value resistors in a marine environment. The largest value I used was 470K for a microphone amp, where PCB leakage was not important. Otherwise, about 100K was as large as I would consider safe. Someone will surely suggest conformal coating. We did some of that but the inability to apply it consistently, even with UV additives, caused problems. Connectors had to be masked, and rework was difficult and messy. No thanks. Incidentally, we also pioneered (with Piezo Technologies) the use of low impedance crystal filters for much the same reasons.

What we did to "solve" the low frequency noise and rumble problem was rather uninspiring. We discovered that 11C44 chips from one country would have a smaller dead band than other. Those were selected for the more critical HF radio synthesizers. It turned out that only one loop was really critical making selecting parts tolerable. Next, the audible noise was removed by high pass filters in the audio circuitry. The trick was keeping the noise below 300 Hz, which proved fairly easy by tweaking the PLL constants. This also had the added benefits of reducing the audible "clunk" whenever the frequency was changed and reducing microphonics[1]. Lastly, the selection of the loudspeaker and the acoustic design of the box, was designed to reduce the low frequencies, mostly to prevent oscillations due to microphonics.

[1] Microphonics are produced by beating on the case, or yelling into the enclosure, which modulates the PLL. This can be heard on the TX and RX audio. In extreme cases, the loudspeaker audio modulates the VCO, causing audio oscillation.
--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

I had an amp that had a microphonic problem. It turned out to be a through hole transistor of a common variety that had a loose element inside and it just happen to be in the preamp section. It was a guitar amplifier head and it would squeal like hell as soon as you turned it up..

Jamie

Reply to
Maynard A. Philbrook Jr.

There are a variety of ways of getting round it. The Motorola way is to have separate up and down outputs, both of which have a nonzero width when the phase error is zero. With that approach, there are two regions where the slope is reduced, but neither of them is where the loop wants to servo. Pulling the loop slightly to one side makes it easier.

If you're relying on the pulse width going to zero to meet your ripple spec, then yes, you're going to have to servo on the flat spot. That's a design error, however.

There's nothing special about 1M. It just has to be enough to pull the servo point a few nanoseconds off the flat spot. If the rest of the loop filter is lower-Z, 100k would probably do it. You just have to design it.

That's pretty much turd-polishing. If you come into a project late, sometimes you have to do that sort of thing, but nobody likes it.

Mechanical sensitivity of oscillators is a pretty well known problem, I agree. Bumping up the loop bandwidth helps a lot.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

I hadn't heard of that method. Good idea and sounds like it would work. I wish I had thought of it. (Note: I was about 25 years old at the time and thought I knew everything there was to know).

Guilty as charged for using the 11C44 in the manufacturers prescribed manner. I wanted to replace it with a double balanced mixer (SBL-1) as a phase detector, but never could get rid of the DC offset or keep it linear. The SBL-1 was chosen mostly because it was already being used in the dual loop synthesizer as a mixer. When I was done with that diversion, patching the problem with high pass filters was the best I could contrive in the time remaining. If I had known about the Motorola trick, I would certainly have tried it.

I just took a closer look at the schematic. The low frequency PLL that supplied the 2nd LO frequency is a 4046 with a 10M resistor pullup. It's an early manual and that was later removed.

Also guilty as charged. Nobody liked it including me. However, the deadline (San Mateo Cow Palace Boat Show) was looming and marketing wanted to demonstrate a working radio. As it turned out, the marketing manager took it with him on an airplane flight to Florida to show some dealers and managed to misplace it for a while at a hotel. When the prototype finally returned to California, it was too late for the boat show. A fair number of people in engineering gave up weekends and evenings for that fiasco and were not amused. As it turned out, I did get another chance to fix things as there were also production problems. However, they were in other areas and the phase detector problems were not addressed.

That helps as long as I don't have to FM modulate the PLL. The PLL should NOT follow the modulation which means that the loop bandwidth has to be lower than the lowest expected modulation frequency (about

300 Hz). The initial design of the synthesizer was not my project, so I don't have any design notes with which to check. I could probably calculate the loop bandwidth from the schematic.

Microphonics is much less of a problem today because of crystal can and SMT VCO's, which are far less susceptible to vibration than the old discrete versions.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

I remember it well. It was fun being that smart, wasn't it? ;)

Back when I was making PLLs for a living, I used to be really fond of the MCL RPD-1 phase detector. It has about 10 dB higher output than a level-7 mixer (2V p-p), and had much lower offset as well. They still make it, or something like it (MPD-1).

I was probably saved from making a very similar mistake by having a very tight noise spec to meet, so that I was pretty sure I couldn't make it work using a PLL chip. (The first thing I tried was a Motorola MC145152, which had that two-output phase detector.) I wound up with a weird fractional-N loop based on a dual modulus prescaler and two decades of synchronous counters, with a chain of two BCD rate multipliers jiggling the carry input to get the bottom two decades. That way I got 8.333... kHz tunability with an 833.3... kHz comparison frequency. (I didn't know about the Digiphase technique, but it would have been under patent still anyway.) The rate multiplier jitter was pretty much out of band, but I still needed to use a VCXO to get good enough close-in phase noise. (The 100-112 MHz output was being multiplied by an additional 120 times to make the LO for a direct broadcast satellite system.)

I really got chucked in the deep end on that one. Fortunately my boss was a noise geek. (Josef Fikart, his name was. Escaped from Czechoslovakia in 1968 on a freight train. Really a good guy.)

I remember grousing about how much reference frequency ripple it caused, and how much nicer a 4046 would be.

The alternative is to put preemphasis on the audio and stuff it into the loop filter along with the phase detector output. That way you can get decent stability as well as good FM.

That's certainly one thing I don't miss about the old days. You just have to remember not to use high-K ceramics in the loop filter, because otherwise the old days are liable to return!

Fun stuff.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.