Frequency counter insensitive

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View


Have just finished assembling one of these,
 http://home.exetel.com.au/marknac/50MHz-Frequency-Meter.htm
the 2003 model Silicon Chip frequency counter. Follow link for complete
article with circuit diagram.

Among other things, it is intended for use as a readout on a signal
generator, an  archaic solid state RCA job from about 1969.
The counter is fine up to about 10MHz but then it runs out of sensitivity
and will no longer read the output of the signal generator.
Problem does not seem to be with the signal generator. Looking at the
waveform on a 'scope, it is putting out around 50mV p-p  all the way up to
20Mhz and there is no drop off in output at the point where frequency
counter ceases to read.
Designer of frequency counter says "Input sensitivity: Typically less than
20mV rms from 1Hz to 100kHz rising to 50mV at 20MHz and 85mV at 50MHz"

I've done some Googling for "frequency counter preamp" and similar search
terms but haven't seen anything very useful.

Any ideas for improving the front end of this counter for more sensitivity?

Peter Howard

Current residents of my crap filter:
soundhaspriority
Robert Morein
George Middius
Phil Allison
Brian Goldsmith



Re: Frequency counter insensitive



Quoted text here. Click to load it

The "preamp" is an MC10116 ECL triple line receiver, and I found one section of
the article amusing.  "the use of a ... LCD ... has several advantages over LED
displays, including much lower current consumption. This allows the unit to be
operated from batteries if required."  Sort of true, but the 10116 current draw
will not be kind to battery operation.

Anyhow, back to your question.  The 10116 is capable of providing a fair bit of
gain, and the last stage should be producing a nice square wave output at all
input signal levels above the threshold.  Stick your CRO onto pin6 (= the base
of Q2) and check the signal shape and swing.  The input biasing of the first
stage of the 10116 is fairly important to get it into the middle of its linear
range, so tweak the bias while watching the result at the output of each line
receiver in turn.  Shift your sig-gen and watch the result - if it isn't showing
a decent swing all the way up to 20 MHz there's something amiss in this area.

I am assuming that you have TRIPLE-checked all component values around the input
and 10116.

The front-end configuration in the schematic has been around for thirty years
with only slight changes.  Even the 2N5485 was used in the 1976 one.  Your
device however adds a bipolar device AND a schmitt gate to get a hard logic
level swing.  The fact that the '76 design drove a logic gate direct from the
10216 (a faster version of the '116) indicates the sort of gain (i.e.
sensitivity) that *should_be* delivered through the line receiver chain.

Re: Frequency counter insensitive



Quoted text here. Click to load it

Rebel, thanks for your advice. I've been fiddling about along the lines you
suggest and decded that I need to concoct a better test signal at a known
level before I can look at expected levels of gain through each 10116 stage.
However, I observe that a sine wave test signal does not end up as a
particularly nice looking square wave at base of Q2 at end of chain. It has
some overshoot and the sides aren't really very vertical.
A square wave test signal starts off clean and ends up looking rather noisy
(jagged on top)
I rechecked all my component values and it's all okay.
I have a spare 10116 coming. They seem to be gettng a bit scarce and
expensive. When it gets here I'll breadboard both it and the original just
to see if the original might be crook. While I'm waiting for it to come I'll
be rigging up a 5 and 10 MHz crystal oscillator with an attenuator.
Anyway, your suggestions were much appreciated because you assured me that
there's nothing unproven or fundamentally wrong with the design and gave me
a plan of attack.

Peter H



Re: Frequency counter insensitive



(snip)

Quoted text here. Click to load it

On re-reading the published article I noted the low stage gain they
quote/use/achieve.  They also would get some effective gain from their common
emitter "level shifter".  But it's a bit "iffy" if the biasing of the level
shifter isn't right for the ECL output signal - that would certainly degrade the
sensitivity.

Testing with a square wave input would be interesting, but in normal service you
would be hoping that RF input signals are low in harmonics so sine sources are
the best for diagnosis.  With an input signal at spec sensitivity, you should be
able to view levels in each stage and work out what is happening.  Keep us
posted.

Also, if you have time for reading it, have a look at:

http://www.onsemi.com/pub/Collateral/AN581-D.PDF

That was the inspiration for using the 10116/10216 in counter front-ends.

Re: Frequency counter insensitive



Quoted text here. Click to load it

Budgie,
Thanks for the link to the application note! I printed it out for bedtime
reading which is where I am headed now.

Tonight I pieced together a little Colpitts osc/buffer/attenuator as signal
source just for this exercise. It's not easy working with a toy two-inch CRO
but a preliminary tryout of the signal source suggests that the 10116 is not
amplifying. I wound the attenuator back while observing sig amplitude on CRO
with one eye and counter display with other eye. The counter won't read
anything less than half a volt at the input terminals. Less than that and
the display starts showing random nonsense. When I get the counter display
*just* locked onto the frequency and follow signal through the stages it
changes from sine wave to a dirty square wave but he signal never gets
bigger. Until I get to collector of Q2 logic level shifter. Lots of voltage
swing there. As you rightly say, the design seems to be getting some
effective gain from that component. Q2 is a PN200 not a BF450 as in the
original design. Kit suppliers say they can't get BF450 so they substituted.

However, unless there's something wrong with my methods I think I have a dud
10116. When you consider the half a volt sensitivity I measure versus  the
stage gains and 20mV sensitivity quoted by designer it all seems to point
that way.

Heigh-ho for Aussie Post and the eventual arrival of another 10116 form
Melbourne! BTW, bloke I deal with on Ebay has 10114's  for about 95c each.
According to spec sheet 10114 is pin-compatible with 10116 and about only
diff I see is that it is just slightly *slower*. In a spirit of enquiry, I
ordered two just to see what happens when I try them.

Peter H




Site Timeline