Attenuator circuit

[Boncuk]

Hi all,

I intend to pep up Space Varmint's frequency counter with an eight digit display and variable gate times from 1ms to 1000s (7 X decade stepping) and a 1/1000 prescaler for frequencies up to 3GHz.

I guess making the counter a way to measure frequencies of low voltage (3.3V chips) and CMOS circuits it will be universal.

Studying attenuator circuits I didn't find exactly what I'm looking for.

The circuit should just be able to reduce (or amplify) the input signal to TTL level at varying inputs of 3 to 12Vpp without much distortion.

Does anybody know of a source for a cheap circuit like the one described?

Boncuk

 

[Nigel Goodwin]

Did I miss the timewarp where we jumped back 30 years?

PIC solutions have been about for almost 20 years, with a VERY old application note that used an OTP PIC and seven segment displays. It was later (still last century) ported to more 'modern' devices and an LCD display - check it out here:



Notice the input stage, a frequency counter isn't a scope, so you don't need (or want) a linear input - you want to clip it and convert it to a square wave. The PIC example (running at a low frequency) makes an auto-ranging 0-50MHz counter with only two chips.

 

[Hero999]

How can a PIC work up to 3GHz?


How about AC coupling the signal and clamping it using diodes?

 

[Nigel Goodwin]

How can a PIC work up to 3GHz?

In EXACTLY the same way as the antique circuit mentioned (and all frequency counters) - you use prescalers to divide it down to something sensible.

How about AC coupling the signal and clamping it using diodes?

You've not checked the PIC circuit have you?

 

[Boncuk]

Did I miss the timewarp where we jumped back 30 years?

PIC solutions have been about for almost 20 years, with a VERY old application note that used an OTP PIC and seven segment displays. It was later (still last century) ported to more 'modern' devices and an LCD display - check it out here:



Notice the input stage, a frequency counter isn't a scope, so you don't need (or want) a linear input - you want to clip it and convert it to a square wave. The PIC example (running at a low frequency) makes an auto-ranging 0-50MHz counter with only two chips.

Hi Nigel,

I saw a lot of junky designs of MCU frequency counters, one of them telling to count to 50MHz at a clock frequency of 20MHz.

I further guess it's not too bad to be able to understand how a frequency works without knowing about programming. In that case the program is wired logic.

Of course I will use a PIC18F2220, not for counting, rather use it for coded outputs for the switching logic - just for the reason that multifunctional interlocking pushbuttons are extremely hard to find nowadays.

Additionally, what should I do with a frequency counter which is not able to count it's own clock frequency?

It might be able to count half of it provided Easter and Christmas eve occur the same day, the sun is shining and the church bells are ringing.

Anyway, none of your sentences answered my question.

Regards

Hans

 

[Nigel Goodwin]

Hi Nigel,

I saw a lot of junky designs of MCU frequency counters, one of them telling to count to 50MHz at a clock frequency of 20MHz.

What's with the ?, a PIC frequency counter counts over 50Mhz even using only a 4MHz clock - don't discount what you don't understand and haven't researched!.

I further guess it's not too bad to be able to understand how a frequency works without knowing about programming. In that case the program is wired logic.

Of course I will use a PIC18F2220, not for counting, rather use it for coded outputs for the switching logic - just for the reason that multifunctional interlocking pushbuttons are extremely hard to find nowadays.

So you'll use a high-end PIC to supplement an old design, rather than a medium range PIC that entirely replaces (and enhances) the old design? - do you not see the lack of logic in that?.

Additionally, what should I do with a frequency counter which is not able to count it's own clock frequency?

It might be able to count half of it provided Easter and Christmas eve occur the same day, the sun is shining and the church bells are ringing.

What frequency counter can't count it's own clock frequency?.

Anyway, none of your sentences answered my question.

So you didn't look at the input circuit of the link I posted ?, which answered your question - even though your question was actually incorrect.

 

[Boncuk]

So you didn't look at the input circuit of the link I posted ?, which answered your question - even though your question was actually incorrect.

I certainly did, but I don't like to search around in an overloaded site full of info about the company which quit producing the kit.

That's a bit off topic, don't you agree?

 

[Nigel Goodwin]

I certainly did, but I don't like to search around in an overloaded site full of info about the company which quit producing the kit.

That's a bit off topic, don't you agree?

No - simply click on the link that says "the circuit" - hardly difficult?.

In case it's too much, I clicked on it for you:

 

[Boncuk]

No - simply click on the link that says "the circuit" - hardly difficult?.

In case it's too much, I clicked on it for you:

Thanks a lot.

BTW, the MPF102 is just good enough for 400MHz - 2.6GHz missing. Would eight MPF102 share the work?

Addionally I guess it is fun designing a high end discrete frequency counter controlled by a microprocessor.

 

[Mike odom]

no... you'd need a whole special type of logic called ECL the use of which I believe to be way above the level of this forum... maybe not, but in 28 years of engineering and an additonal 10 years of practicing electronics, I've never used them.

 

[Hero999]

Modern high speed CMOS logic can work at 3GHz, why would you use ECL?

 

[Nigel Goodwin]

Thanks a lot.

BTW, the MPF102 is just good enough for 400MHz - 2.6GHz missing. Would eight MPF102 share the work?

Addionally I guess it is fun designing a high end discrete frequency counter controlled by a microprocessor.

As you already know, it's a 50MHz counter - and you're not going to get front ends for GHz frequencies, you go directly in the ECL dividers.

 

[Nigel Goodwin]

Modern high speed CMOS logic can work at 3GHz, why would you use ECL?

What type numbers of CMOS dividers work at 3GHz? - as far as I'm aware ECL was the only suitable technology (and pretty rare now).

 

[Hero999]

My computer works at 3GHz, the CPU is CMOS based, not ECL.

Here's a paper describing a 3GHx CMOS frequency divider.
https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.12.5856&rep=rep1&type=pdf

 

[Boncuk]

Hi Nigel,

I'm going to try out that one (attached) hoping it will work alright.

For the implemented functions a PIC16F84 won't do - not enough I/Os.

To switch functions a PIC18F2220 is completely busy, including RA7 (OSC1)

Nothing against religions, mine is "wired logic". (which I love!)

And nothing against Microchip or ATMEL. Their products are really great, but lead to the temptation to forget about producing a real populated PCB.

Till now the IC count is 27, not including the 1/1000 prescaler (for easy math). The counter itself must be able to handle 3MHz at 3GHz input. (laughably low frequency)

As mentioned before, design is fun for me, and if anybody in this (outstandingly excellent forum) has the time and drive to build the circuit (after testing and publishing solely at ETO)
it will give me more satisfaction than 1,000.000.000 US$ on my bank account. (unable to spend till the end of my life).

I hope you got me this time

Regards

Hans

 

[ericgibbs]

hi Hans,
Look at this link:
Silicon Chip Online - UHF Prescaler For Frequency Counters

 

[Boncuk]

As you already know, it's a 50MHz counter - and you're not going to get front ends for GHz frequencies, you go directly in the ECL dividers.

With other words, you didn't answer my question correctly and logically.

No problem, I can live with crap.

 

[Nigel Goodwin]

My computer works at 3GHz, the CPU is CMOS based, not ECL.

Nothing whatsoever to do with CMOS logic chips

TTL and CMOS are basically limited to about 50MHz, which is why counters (both logic and micro-controller) have that sort of range - for higher than that you use ECL dividers in front of it, and basically 75/50 ohm matched impedances. My counter at work uses different BNC sockets, 1Mohm for low ranges (like a scope) and 50 ohms for the highest range (200Mhz)

 

[Space Varmint]

Did I miss the timewarp where we jumped back 30 years?

PIC solutions have been about for almost 20 years, with a VERY old application note that used an OTP PIC and seven segment displays. It was later (still last century) ported to more 'modern' devices and an LCD display - check it out here:



Notice the input stage, a frequency counter isn't a scope, so you don't need (or want) a linear input - you want to clip it and convert it to a square wave. The PIC example (running at a low frequency) makes an auto-ranging 0-50MHz counter with only two chips.

Perhaps you missed the point of my post. That one in your link uses HUGE I mean gigantic amounts of power. Those dot matrix diplay will eat a battery in no time and a processor running 4MHz is hungry too. Mine was posted as a frequency counter that works and is low power and simple. Now if he had done his with a much lower crystal frequency and used an LCD display (7 segment) then it would be more impressive. Let's face it. In todays electronics we are competing with cheaply made Chinese junk. And low power is the name of the game.

 

[Space Varmint]

They use ECL in Cray computers. That's one reason they are so fast. ECL (Emitter Coupled Logic) is some form of common collector logic. Transistors switch much faster in emitter follower configuration. I use ECL in PLL's and freq counters. Let's say I need a PLL to operate at 1GHz. I could run the VCO at that frequency much easier than using frequency multipliers which have a poor range in general. They tend to double, or halve with small changes in phase. So if the the VCO runs at 1GHz then I have to divide it down to a point where I can work with the logic. If I want 1KHz resolution then I divide it again with TTL but the TTL is useless until I get it down to the 50MHz Nigel stated.