OT: Electronic Design!

In a surprise change from the usual topics discussed in this group, I'm posting a question about electronic design!

Win kindly reposted this enhanced schematic of an amplifier board I've been troubleshooting:-

formatting link

My question relates to the voltage-to-current section which comprises 6 transistors plus various passive 'support' components. My previous experience is confined to discrete amplifier stages where each transistor is dc-isolated from its neighbours by coupling capacitors and consequently have easily determined biasing arrangements for each stage. This is very different as all 6 transistors are directly inter-related and inter-dependent (for example the collector output voltage(s) of one has to be compatible with the base input of the next and so on. Not only this, but their emitters all share series connections with their compliments. This kind of arrangement is a PITA to troubleshoot, as a problem with just one active device (or its 'supporting components') causes weird voltage readings among the other 5 as well. Anyway, it occurred to me it must have been an even bigger PITA to actually *design* such a beast in the first place, given all the inter- dependence of the bias and signal voltages which all have to be accommodated. What thought processes would the designer of this board have had to go through in order to come up with what is effectively a monolithic 6- transistor "stage" with all the multiple complications that go with it? Where does one even begin??

--
This message may be freely reproduced without limit or charge only via  
the Usenet protocol. Reproduction in whole or part through other  
 Click to see the full signature
Reply to
Cursitor Doom
Loading thread data ...

Should you be posting on topic stuff?

Reply to
tabbypurr

*Long* term contributors here (20 years and more) will know I do *sometimes* post on-topic stuff. I'm just a hobbyist. I'll never be able to design stuff like this, but am fascinated by the approach real designers take with certain challenges that I couldn't even begin to get to grips with.
--
This message may be freely reproduced without limit or charge only via  
the Usenet protocol. Reproduction in whole or part through other  
 Click to see the full signature
Reply to
Cursitor Doom

So it's a V - I converter/transconductance amplifier. The design choice to use a cascade of three is likely dictated by knowing that for whatever purpose, you need an overall transconductance (voltage in -> current out) gain of a given amount at a certain max operating frequency.

You can increase the transistor bias current to increase the current gain but all active devices have a gain-bandwith product / "fT" so if you increase it too much to get the gain you need you may no longer have the bandwidth you need.

So I suppose they ran the math on how many stages they'd need to get the current gain but keep a wide enough bandwidth to not roll off the bandwidth of interest too bad, using the active devices that were available to use or at the price they wanted to pay for 'em. And came up with three stages.

The decision to multi-stage the small-signal amp and then all the other support components required to bias up a multi-stage differential amplifier into the operating area it works best falls out of knowing how much current gain you need at what max frequency, that you need an amp that rejects common-mode, and what the performance constraints of the active devices in your budget are (or could be attained at reasonable cost.)

It turned out two stages is too few for performance requirements, and four is too much money. Yeah the passives were all selected such that the DC operating point of everything all falls into place at quiescent very nicely. DC biasing BJT networks is pretty easy if they're all assumed to be in the active region, the base is just assumed to be 0.6 volts above or below the emitter depending on NPN/PNP you can just about do it by inspection.

They were also selected so the AC performance is about right too but an adjustment trimmer was also provided to tweak that up in the AC feedback path of the final stage, R26 and R27 don't do anything at DC.

Reply to
bitrex

For a given gain, sure. But for any active device of a given type your max potential gain/bandwidth ratio is a constant determined by the physical characteristics of the device. It applies for op amps and it applies for discrete transistors, too.

Reply to
bitrex

I was quite surprised when I scoped the signal path of this section as after each transistor, going from source to load, the signal level halves. I'd have expected this if they were using transformers to bump up the current, but not with active devices. If I can't find the cause of the fault I might just bypass these 6 transistors altogether and replace them with a pair of op amps. Did they not have op amps in 1977? ;-)

--
This message may be freely reproduced without limit or charge only via  
the Usenet protocol. Reproduction in whole or part through other  
 Click to see the full signature
Reply to
Cursitor Doom

It has been over 40 years since i was "into" such things, but as long as your transistor emitter current did not materially increase Rbb', increasing the operating current would increase Ft.

Reply to
Robert Baer

Yes but they were bad and your idea is bad.

Reply to
bitrex

But can probably keep the four HV parts.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

--------------------

** You have posted a pic of a weird looking dog and ask us to unravel its pedigree. 100% absurd.

Like with the dog, one has to know it's ancestry first in order to explain why it looks the way it does.

Do you even know where that schem comes from at all?

What commercial product ?

So many scopes have been designed over the decades and they all have some similarities - designers are terrible plagiarisers.

Products from the same maker and period likely have whole sections copied from earlier ones or copied from similar products from other makers.

Sometimes with improvements, but more likely with fewer and cheaper parts to save pennies.

Really, only the actual designer can explain how he arrived at a particular circuit - and none of them are talking.

.... Phil

Reply to
Phil Allison

I think (when it works) that V -> I -> V topology helps simplify making the wide-band differential HV signal to drive the CRT quite a bit. You get a 150 volt P2P swing out of a four-transistor TIA. gain is approximately Iin * C10||R29. Bypassing bias resistor R35(?) makes sure both output transistors can get enough AC drive current. Idk if there's a name for that topology it's a little like a White cathode follower, but not.

Reply to
bitrex

Particularly since all the loop gain is in the output stage and output is taken from the collectors rather than emitters.

Reply to
bitrex

HP 8565A, is in the dropbox URL he posted my dude

Reply to
bitrex

But it's on topic, that's the problem.

Reply to
tabbypurr

Not exactly. The problem is the usual one - he hasn't said where it came from. or what it was actually intended to do, as Phil Allison has pointed out

This does get mentioned from time to time when people post ostensibly on-topic items, but with way too little context.

It's not unusual for a crucial detail to only emerge after a dozen or so posts have beaten around quite a different bush from the one the OP wanted beaten.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

Aren't we talking about the horizontal deflection amplifier for the CRT in HP 8565A spectrum analyzers, c. 1977? See:

formatting link

--
 Thanks, 
    - Win
Reply to
Winfield Hill

The string "8565" didn't trigger anything in my memory. Cursitor Doom just talked about an "amplifier board".

The circuit diagram does mention that it's driving a horizontal CRT deflect ion plate. Somebody who knew a bit about HP 8565 might be able to work out how fast it might be going, but that's more digging that I'd be bothered do ing for Cursitor Doom.

The fact that it's an HP part suggests that it probably worked when it was new, but it's no guarantee that it was well-designed. Some of their stuff t hat I did have to dig into wasn't impressive.

My favourite American botch job was a Princeton Applied Research boxcar int egrator that we got stuck with when we had to copy a Siemens development pr ototype for Thompson CSF. The spec sheet said that it could sample at up to 5MHz, but when we checked it died at 2.3MHz. I could get it to go faster b y sticking in a 74H part, but that only got it up to 2.7MHz.

When we contacted PAR, they agreed that it didn't meet it's spec.

Not its only defect - but a particularly blatant one.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

The HP design is fine. All these aspects were discussed in a thread started 13 Oct, Title: Amplifer Design Message-ID:

--
 Thanks, 
    - Win
Reply to
Winfield Hill

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.