Forums

Amplifer Design

Started by Cursitor Doom October 13, 2019
Cursitor Doom wrote:
> On Mon, 14 Oct 2019 05:59:23 -0700, Winfield Hill wrote: > >> There's a 236MB file of the HP 8565A Operating and Service Manual, >> dated 1977, 424 pages, on the web. Unfortunately, it's locked, so I >> could not extract the pdf schematic page. But I expanded it, and made >> an image of the HV output amp. >> >> https://www.dropbox.com/s/o4tybr81cefbk5n/8565A_amp.GIF?dl=1 > > And a splendid image it is too. You're obviously way better than me at > enhancement. > >> The 1st diff-amp stage has 562R emitter and 2k15 collector resistors >> for G=3.8. The 2nd stage has 750 + 500-ohm trim emitter resistors and >> a current output. The HV output stage has a 19k6 feedback resistor, >> for G = 15 to 26, 19 nominal. A 1pF feedback cap would set the video >> bandwidth at 8MHz. >> I imagine Zout is less than 1k. Push-pull PNP & NPN BJTs, running at >> 7mA, and dissipating 1 watt, go from 2V to 140V. There are two >> complementary amps, for two deflection plates. > > It seems odd to me that there's no minus 158V supply for the final > voltage amp. It looks like they're relying solely on attracting the > electron beam from side to side with positive voltages swapping back and > forth between plates rather than a combination of attraction and > repulsion as I would have intuitively expected. :-/
The electrons see only the field, not the potential. Jeroen Belleman
Cursitor Doom wrote...
> > On Mon, 14 Oct 2019 05:59:23 -0700, Winfield Hill wrote: > >> There's a 236MB file of the HP 8565A Operating and Service Manual, >> dated 1977, 424 pages, on the web. Unfortunately, it's locked, so I >> could not extract the pdf schematic page. But I expanded it, and made >> an image of the HV output amp. >> >> https://www.dropbox.com/s/o4tybr81cefbk5n/8565A_amp.GIF?dl=1 > > And a splendid image it is too. You're obviously way better > than me at enhancement.
The credit belongs to the person who scanned of the manual. The schematic is on one of those super-wide foldout pages, which have to be scanned in sections and re-combined. The pdf's quality is as good or better than my screen capture.
>> The 1st diff-amp stage has 562R emitter and 2k15 collector resistors >> for G=3.8. The 2nd stage has 750 + 500-ohm trim emitter resistors and >> a current output. The HV output stage has a 19k6 feedback resistor, >> for G = 15 to 26, 19 nominal. A 1pF feedback cap would set the video >> bandwidth at 8MHz. >> I imagine Zout is less than 1k. Push-pull PNP & NPN BJTs, running at >> 7mA, and dissipating 1 watt, go from 2V to 140V. There are two >> complementary amps, for two deflection plates. > > It seems odd to me that there's no minus 158V supply for the > final voltage amp.
The differential current into the output stage summing junction is -3.6mA, so with the 19k6 feedback resistor, the nominal output voltage on each side is +70 volts. As one side swings up by +50V to +120 volts, the other side swings down by -50V to +20 volts. The deflection electrodes see a nominal 0 volts, and swing from +100 volts to -100 volts, more or less, set by the gain trimpot. -- Thanks, - Win
Winfield Hill wrote...
> >Cursitor Doom wrote... >> >> On Mon, 14 Oct 2019 05:59:23 -0700, Winfield Hill wrote: >> >>> There's a 236MB file of the HP 8565A Operating and Service Manual, >>> dated 1977, 424 pages, on the web. Unfortunately, it's locked, so I >>> could not extract the pdf schematic page. But I expanded it, and made >>> an image of the HV output amp. >>> >>> https://www.dropbox.com/s/o4tybr81cefbk5n/8565A_amp.GIF?dl=1 >> >> And a splendid image it is too. You're obviously way better >> than me at enhancement. > > The credit belongs to the person who scanned of the manual. > The schematic is on one of those super-wide foldout pages, > which have to be scanned in sections and re-combined. The > pdf's quality is as good or better than my screen capture. > >>> The 1st diff-amp stage has 562R emitter and 2k15 collector resistors >>> for G=3.8. The 2nd stage has 750 + 500-ohm trim emitter resistors and >>> a current output. The HV output stage has a 19k6 feedback resistor, >>> for G = 15 to 26, 19 nominal. A 1pF feedback cap would set the video >>> bandwidth at 8MHz. >>> I imagine Zout is less than 1k. Push-pull PNP & NPN BJTs, running >>> at 7mA, and dissipating 1 watt, go from 2V to 140V. There are two >>> complementary amps, for two deflection plates. >> >> It seems odd to me that there's no minus 158V supply for the >> final voltage amp. > > The differential current into the output stage summing junction > is -3.6mA, so with the 19k6 feedback resistor, the nominal output > voltage on each side is +70 volts. As one side swings up by +50V > to +120 volts, the other side swings down by -50V to +20 volts. > The deflection electrodes see a nominal 0 volts, and swing from > +100 volts to -100 volts, more or less, set by the gain trimpot.
That's a net peak-peak deflection-voltage swing (difference) of 200 volts, or nearly 300 Vpp at the amplifier's extreme limits. The 7mA current output capability of the amplifier means, if the plate capacitive load, plus the two BJT's Cob, is say 30pF, that the maximum output slew rate is S=i/C = 230V/us. Therefore it can slew the p-p voltage of 200V in under 1 us. That's a pretty respectable performance for a simple, low-cost circuit in 1977. -- Thanks, - Win
On Mon, 14 Oct 2019 07:59:44 -0700, Winfield Hill wrote:

> The differential current into the output stage summing junction is > -3.6mA, so with the 19k6 feedback resistor, the nominal output voltage > on each side is +70 volts. As one side swings up by +50V to +120 > volts, the other side swings down by -50V to +20 volts. The deflection > electrodes see a nominal 0 volts, and swing from +100 volts to -100 > volts, more or less, set by the gain trimpot.
OK, many thanks for that. Whilst I'm absorbing your analysis, I've discovered a problem with Q8. The base-emitter and base-collector voltages are both around -0.68V and static. OTOH, Q8's compliment Q13 has base-emitter voltage of 0.5 - 0.7V cycling steadily in line with the timebase as it should and a healthy base-collector voltage of ~11V. Clearly Q8 can't operate with those voltages and it looks like it's got a significant C-E internal short. Can't say for certain without hooking it out of circuit, but there's not much else in that area it could be! -- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
On Mon, 14 Oct 2019 08:45:59 -0700, Winfield Hill wrote:

> That's a net peak-peak deflection-voltage swing (difference) of 200 > volts, or nearly 300 Vpp at the amplifier's extreme limits. The 7mA > current output capability of the amplifier means, if the plate > capacitive load, plus the two BJT's Cob, is say 30pF, that the maximum > output slew rate is S=i/C = 230V/us. Therefore it can slew the p-p > voltage of 200V in under 1 us. That's a pretty respectable performance > for a simple, low-cost circuit in 1977.
Indeed. In fact I think IIRC I read somewhere that CRTs beat liquid crystal hands down in this regard - although that's more to do with the superb agility of an electron beam I would imagine. -- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
Winfield Hill wrote...
> > Winfield Hill wrote... >>Cursitor Doom wrote... >>> On Mon, 14 Oct 2019 05:59:23 -0700, Winfield Hill wrote: >>> >>>> There's a 236MB file of the HP 8565A Operating and Service Manual, >>>> dated 1977, 424 pages, on the web. Unfortunately, it's locked, so I >>>> could not extract the pdf schematic page. But I expanded it, and made >>>> an image of the HV output amp. >>>> >>>> https://www.dropbox.com/s/o4tybr81cefbk5n/8565A_amp.GIF?dl=1 >> >>>> The 1st diff-amp stage has 562R emitter and 2k15 collector resistors >>>> for G=3.8. The 2nd stage has 750 + 500-ohm trim emitter resistors and >>>> a current output. The HV output stage has a 19k6 feedback resistor, >>>> for G = 15 to 26, 19 nominal. A 1pF feedback cap would set the video >>>> bandwidth at 8MHz. >>>> I imagine Zout is less than 1k. Push-pull PNP & NPN BJTs, running >>>> at 7mA, and dissipating 1 watt, go from 2V to 140V. There are two >>>> complementary amps, for two deflection plates. >>> >>> It seems odd to me that there's no minus 158V supply for the >>> final voltage amp. >> >> The differential current into the output stage summing junction >> is -3.6mA, so with the 19k6 feedback resistor, the nominal output >> voltage on each side is +70 volts. As one side swings up by +50V >> to +120 volts, the other side swings down by -50V to +20 volts. >> The deflection electrodes see a nominal 0 volts, and swing from >> +100 volts to -100 volts, more or less, set by the gain trimpot. > > That's a net peak-peak deflection-voltage swing (difference) of > 200 volts, or nearly 300 Vpp at the amplifier's extreme limits. > The 7mA current output capability of the amplifier means, if the > plate capacitive load, plus the two BJT's Cob, is say 30pF, that > the maximum output slew rate is S=i/C = 230V/us. Therefore it > can slew the p-p voltage of 200V in under 1 us. That's a pretty > respectable performance for a simple, low-cost circuit in 1977.
The video output-stage's modest 1-watt class-A power dissipation is nominally divided between the two BJT's, but a fault situation could force most of it to be in one part. That's too much power for a modern TO-92 or SOT-23 type, but the metal-can parts in the old days had no problem. Furthermore, a clip-on heat sink could easily be added. HP affixed rather large ones to the TO-39 parts. HP only gives HP numbers, but it's safe to say the manufacturer's original is no longer available. Now I struggle with high-voltage SOT-223 or TO-126 parts for my HV amplifiers, if I can find them. -- Thanks, - Win
On a sunny day (Mon, 14 Oct 2019 15:55:55 -0000 (UTC)) it happened Cursitor
Doom <curd@notformail.com> wrote in <qo25qa$kaj$2@dont-email.me>:

>On Mon, 14 Oct 2019 08:45:59 -0700, Winfield Hill wrote: > >> That's a net peak-peak deflection-voltage swing (difference) of 200 >> volts, or nearly 300 Vpp at the amplifier's extreme limits. The 7mA >> current output capability of the amplifier means, if the plate >> capacitive load, plus the two BJT's Cob, is say 30pF, that the maximum >> output slew rate is S=i/C = 230V/us. Therefore it can slew the p-p >> voltage of 200V in under 1 us. That's a pretty respectable performance >> for a simple, low-cost circuit in 1977. > >Indeed. In fact I think IIRC I read somewhere that CRTs beat liquid >crystal hands down in this regard - although that's more to do with the >superb agility of an electron beam I would imagine.
In fact it is not very good for those years Look up a Tek from those days, One big difference is that the modern tubes in those days had a much higher sensitivity, Maybe 20 Vpp or so for FSD. I build a 300 MHz scope in 1976 or there about using a circuit diagram that somebody at Tek (accidently) published, met the guy later, he got headwind for that nice circuit though. I used an east European tube. As to LCD, that works totally different: Signal is digitized (at what ever speed) put into a memory, and that memory can be read out and put in the LCD display's memory. the LCD controller reads that LCD memory at its own (usually much slower) speed and refreshes the LCD display. (even if there is no more signal on the scope input), this gives you already 'storage'. Also you can that do signal processing such as Fourier transform on the data in that memory. That is also what I do with scope_pic, there only 256 bytes of memory available in the PIC.... http://panteltje.com/panteltje/pic/scope_pic/index.html But it is also the reason digital scopes lie. With digitizing comes aliasing and all sorts of unwanted effects. I hold on to my old 10 MHz Trio analog scope :-) With a bit of clever use it is all I need (and I work with GHz stuff) heterodyne (down-mix), or use a RTL_SDR USB stick as spectrum analyzer (<30 $): http://panteltje.com/panteltje/xpsa/index.html Looking at signals 2.4 GHz with a down mixer is then also no problem Or without down-mixer at around 1.5 GHz in real time. Even decoded GPS with that: http://michelebavaro.blogspot.com/2012/04/spring-news-in-gnss-and-sdr-domain.html scroll all the way down to my entry. Good stuff does not need to cost much, most of those 'boat-anchors' are not even good at anchoring... Spectrum often tells you a lot. As to diff amplifiers: The output depends on the difference_ between the inputs, so if one input is at a fixed voltage and the other changes, then there is a difference between the inputs created. If both inputs move the same way wit the same amplitude there is no difference (and no output) All within a reasonable range between the supply rails of course, or at least it should be.
On Mon, 14 Oct 2019 09:20:46 -0700, Winfield Hill wrote:

> The video output-stage's modest 1-watt class-A power dissipation is > nominally divided between the two BJT's, but a fault situation could > force most of it to be in one part. That's too much power for a modern > TO-92 or SOT-23 type, but the metal-can parts in the old days had no > problem. Furthermore, a clip-on heat sink could easily be added. HP > affixed rather large ones to the TO-39 parts.
Indeed they did. Here's the board in question and you can't miss the four final transistors with their rather generously-proportioned heat sinks: https://yandex.com/collections/card/5da4bb75ada5960091318610/ This is the kind of layout I like: no smds and plenty of space between the components to probe connections. I can't do that with anything built after SMDs came in, I lack the fine motor skills required.
> HP only gives HP numbers, but it's safe to say the manufacturer's > original is no longer available. Now I struggle with high-voltage > SOT-223 or TO-126 parts for my HV amplifiers, if I can find them.
They have an odd practice of quoting either one or the other. So the first four TO-18 tin can transistors are shown as 2N3251 devices, but the next stage uses "4-404 826" BJTs and goodness knows what they are. I'm not even going to bother to stick that into Google because I know it won't throw up anything helpful at all. But on the plus side, I know enough about the functioning of the circuit now to be able to drop a couple of closely-matched jellybeans in there in place of Q8 and Q13 and in this x-amp at any rate, given its modest task, will quickly be back in business again. All this old vintage stuff remains so readily serviceable. :-) -- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
On Mon, 14 Oct 2019 17:16:12 +0000, Jan Panteltje wrote:

> I hold on to my old 10 MHz Trio analog scope :-) > With a bit of clever use it is all I need (and I work with GHz stuff) > heterodyne (down-mix), or use a RTL_SDR USB stick as spectrum analyzer > (<30 $): > http://panteltje.com/panteltje/xpsa/index.html > Looking at signals 2.4 GHz with a down mixer is then also no problem Or > without down-mixer at around 1.5 GHz in real time.
I really, really admire your resourcefulness, Jan. You clearly have an excellent grasp of your subject. However, why make life hard for yourself? I mean a 10Mhz scope and mixing down??? If you're working with GHz stuff then surely you can afford something a lot better than that! I had a Belgian friend once many years ago and he frequently said the Dutch were "careful with their money" (shall I say it tactfully) and here you are now proving it! :-D -- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
On a sunny day (Mon, 14 Oct 2019 18:31:55 -0000 (UTC)) it happened Cursitor
Doom <curd@notformail.com> wrote in <qo2eur$4sq$2@dont-email.me>:

>On Mon, 14 Oct 2019 17:16:12 +0000, Jan Panteltje wrote: > >> I hold on to my old 10 MHz Trio analog scope :-) >> With a bit of clever use it is all I need (and I work with GHz stuff) >> heterodyne (down-mix), or use a RTL_SDR USB stick as spectrum analyzer >> (<30 $): >> http://panteltje.com/panteltje/xpsa/index.html >> Looking at signals 2.4 GHz with a down mixer is then also no problem Or >> without down-mixer at around 1.5 GHz in real time. > >I really, really admire your resourcefulness, Jan. You clearly have an >excellent grasp of your subject. However, why make life hard for >yourself? I mean a 10Mhz scope and mixing down??? If you're working with >GHz stuff then surely you can afford something a lot better than that! I >had a Belgian friend once many years ago and he frequently said the Dutch >were "careful with their money" (shall I say it tactfully) and here you >are now proving it! :-D
Na... farmers are revolving here against being forced by the government to cut there herd to save on CO2.. They occupied gov buildings and got what they wanted... https://www.nu.nl/binnenland/6004031/boeren-forceren-deur-provinciehuis-groningen-tijdens-stikstofprotest.html I was not referring to mixing down to the 10 MHz. I was referring to mixing down 2.4 GHz to 1.5 GHz or so for use with an RTL_SDR USB stick as spectrum analyzer, that I did. OTOH if you scope the 470 kHz or 9 MHz IF of a shortwave radio with a 10 MHz scope then you can see a lot about the modulation. It is the experimenter, not the instrument. I do remember .. well getting big boat_anchors in your shop may impress potential customers :-) You never say what you want to do with your 'instrument' if you are a collector OK... My suggestion to you is: design one. Start designing a simple scope, look at how others did it, take it from there. So, as to GHz: https://www.ebay.com/itm/272411458376 much more fun to play with than old boat_anchors. that is what my xpsa spectrum analyzer software runs on. Plus a hundred other things, AIS, planes, radio, TV. etc etc. High speed IQ sampling. 1 pmm... frequency measurement, what not. Build something.