Oscilloscope under construction.

[GK]

Six transistors and two diodes. I'm currently assembling the (deflection) amplifier board.



The preliminary final schematic:



The P.U.T. injection-locked horizontal timebase/sweep-generator synchronised to a sinewave:



This is the closest semiconductor equivalent to a 884 thyratron sweep circuit that I could come up with.

The inspiration:
   

[tautech]

Fascinating as always from you Glen. 

Clever use of a PUT. 

Expected BW ?

[zapta]

Expected BW ?

AC only, right?

[GK]

Fascinating as always from you Glen. 

Clever use of a PUT. 

Expected BW ?

About 100kHz or so.

The amplifier board with the long-tail pairs is actually my second deflection amplifier design. The original board had a pair of single-ended amplifiers driving only one deflection plate of each pair as per the Philips TA115. That made this a 4-transistor 'scope. My original aim was to replicate the TA115 in solid state (3 transistors). However the CRT I choose, as it transpires, doesn't work well with single-ended deflection plate drive. The deflection plates act to a degree as electron beam accelerators and if the mean voltage between a deflection plate pair does not remain constant with ac signal swing (as it does with differential drive) the net electron beam acceleration potential will be modulated in accordance. Deflection sensitivity is inversely proportional to the acceleration potential, so the single-ended drive results in significant deflection non-linearity - I was getting about 25%. Imagine an AM envelope display of an unmodulated carrier. The vertical deflection sensitivity at the left-most extremity of the CRTs display was ~25% less than at the right most, so the displayed envelope became trapezoidal shaped. I liked the ultra simplicity, but I decided that I couldn't live with such horrible non-linearity, so I re-designed the amplifier board for differential drive and the total transistor count grew by two to six.
I also couldn't make do without the emitter-follower buffer for the sweep generator. The capacitor charging current must be kept below (by a safe margin) the PUTs valley current, otherwise the PUT, once triggered, won't turn off again. The input impedance of the BJT horizontal deflection amplifier could not be (practically) made high enough to make loading of the capacitor charging circuit acceptable. So I'm afraid that, as far as the active device count goes in this application, tubes still reign supreme.

Also, unfortunately, the PUT appears to have gone the way of the tunnel diode. AFAIK the last manufacturer of the 2N6027 was ON Semi and they've now stopped production. I got five examples from a Jaycar store which still had stock.

[Cyberdragon]

Also, unfortunately, the PUT appears to have gone the way of the tunnel diode. AFAIK the last manufacturer of the 2N6027 was ON Semi and they've now stopped production. I got five examples from a Jaycar store which still had stock.

What you talking about willis?! We don't need ON Semiconductors!
http://www.intertexelectronics.com/Transistor-2N6027-Unijunction-P11164.aspx
http://www.talonix.com/shop/item.aspx?itemid=3467

[GK]

Those are just distributors with stock, not manufacturers. Though Central Semiconductor Corp is still actively producing the 6027/6028, so it hasn't quite gone the way of the tunnel diode just yet:

https://www.centralsemi.com/paraSearch/parametricSearch_v1_1.php
Listed under "thyristors".

[David Hess]

And Mouser carries the Central Semiconductor 2N6027 and 2N6028 PUTs at a reasonable price.

[ali6x944]

Good job GK, this is fascinating...
How with a hand full of parts u could build a working scope,
Man this is just fascinating.... keep it up m8 I wish u best of luck

[StillTrying]

In a junk box, I've still got the one I built at about 14 years old, as I had no idea what I was doing it's unique!

[Berni]

Neat oldschool project there. 

Its quite amazing how much can be done with only a few transistors. Today nobody cares since semiconductors are as cheep as passives unless its a fancy special part of a fancy IC.

[oldway]

I think the real fun would be to make a true TA155 with 902 crt, 6J7's and 884 thyratron.

The dg7-32 crt is more modern and not intended to work with asymetric plate drive.
It is also able to work up to about 3Mhz and it is a pity to use it for only 100 Khz bandwith.

Why not make a 3Mhz project ?
Here, some schematics of a simple MBLE oscilloscope BEM16...

[GK]

I think the real fun would be to make a true TA155 with 902 crt, 6J7's and 884 thyratron.

The dg7-32 crt is more modern and not intended to work with asymetric plate drive.

I own and have refurbished my TA155 and the 902's deflection non-linearity really isn't that far behind the DG7-32 with only one plate of each pair driven. I have over thirty NOS electrostatic CRTs ranging 1" to 5" dia. including half a dozen DG7-32s. This isn't the first scope project and it won't be the last.

 

[GK]

I got the amplifier board installed and mostly wired this evening. I'll finish it off tomorrow or the next evening.

[Circlotron]

The appearance reminds me of this one that I built many moons ago.

[GK]

The appearance reminds me of this one that I built many moons ago.

Hmm, don't have that issue of RT&H but I think I know the design. Does it use four 6BX6 pentodes in the vertical amplifier? 

[GK]

Soldered the final connection this evening. Now all I have to do is finish cable-tying and drill holes for half a dozen P-clamps to secure/tidy up the rest of the wiring. But anyway, its electrically finished and 100% operational.

[GK]

I could bang on about just how good the two-transistor injection-locked linear sweep generator (time base) works in practice, but I think a video is better. I used a triangle wave for the vertical input signal to better show the deflection/display/sweep linearity. The time base is synced predominantly by adjusting the time base coarse and fine frequency controls so that the timebase frequency is an exact(ish) divisible fraction of the input signal frequency. The time base will also lock to signals that are lower in frequency (half, quarter, etc). 

I challenge anyone to design a significantly simpler solid-state time base!   

http://youtu.be/ZSPjTfhK25A

[setq]

Interesting thread and nice work!

I've got a long term plan to build a more modern clone of the Textronix 321A (as described here: http://www.helo.de/helo/tek/portable/321a/321a_e.htm ). It was one of the first purely solid state scopes and only contains a few transistors which roughly replaced similar functions as the previous revisions with vacuum tubes. That managed a good 6MHz bandwidth and I reckon you can do better with newer parts.

Replying to the above, Tektronix produced miller integrator based timebases. Check the SM for the Tek 321A. It's a very nice design.

[rstofer]

That is a magnificent project.  I built something much less capable back in the late '50s from plans in the ARRL Handbook.  MUCH less capable but with vacuum tubes...

I would certainly be interested in how the front panel was built.  It seems that panels are the bane of all hobby projects.  So, I'll share mine...

The panel is two layers of clear acrylic sheet with a printed sheet sandwiched between:

ETA:  Having to drill so many precisely spaced holes was the motivation for building a CNC mill.  Kind of a twofer project.

[ajb]

 

Do you buy or make that aluminum bar with the tapped holes you use to frame the chassis?  That'd be handy to have around, if it can be bought cheaply ready-made. 

[David Hess]

ETA:  Having to drill so many precisely spaced holes was the motivation for building a CNC mill.  Kind of a twofer project.

My solution for that has been to print out a 1:1 punch out diagram, tape it to the front panel, and use an automatic center punch before drilling.

[TheBay]

Amazing bit of work there, nothing better than seeing something you built do something 

[timb]

ETA:  Having to drill so many precisely spaced holes was the motivation for building a CNC mill.  Kind of a twofer project.

My solution for that has been to print out a 1:1 punch out diagram, tape it to the front panel, and use an automatic center punch before drilling.

I use inkjet transparency film for this. Print the design out, tape it on and use it as a guide on the drill press.

[bitseeker]

@GK: Awesome scope and amazing how few components were needed.

@rstofer: That's a lot of holes to drill. The acrylic sandwich looks great.

[timb]

ETA:  Having to drill so many precisely spaced holes was the motivation for building a CNC mill.  Kind of a twofer project.

My solution for that has been to print out a 1:1 punch out diagram, tape it to the front panel, and use an automatic center punch before drilling.

I use inkjet transparency film for this. Print the design out, tape it on and use it as a guide on the drill press.