Hugo Holden
Veteran Member
I bought a spare VDU for restoration for my PET and some NOS CRT's to test. I'm writing a detailed article on this VDU restoration and the LOPT. Including how to deal with the rust and re-paint the cabinet. Since I had it, it gave me a good opportunity to fully investigate the design and make some modifications to improve it.
The article is very large and not finished, because of the extensive work on the LOPT analysis (directed toward using substitute LOPT's if needed) so I thought I would post some of the other work more briefly. The idea was to make the modifications reversible with as little change to the pcb as possible.The things that require improving are:
1) Poor Horizontal scan linearity - reason, no magnetic linearity inductor- Solution add high speed high voltage germanium rectifier and alter S correction capacitor value to 8uF.
2) Poor focus- reason, inadequate focus voltage for G4 electrode, needs to be adjustable around 330V. Only GND & 85V available inside VDU - solution; add transformer/rectifier/focus control powered by 580V.
3) Inadequate brightness control range (both VDU's same problem)- reason, the manufacturer's "cheated" with the design of the video output stage because the voltage ratings of the chosen output transistors were not high enough for normal CRT cathode voltages around 70v and also they did not have a higher first anode voltage than 85V (ideally for the CRT ist anode its 120V if the cathode voltage was higher around 70V) so they were forced to regulate the cathode voltage down with a zener diode. This required a higher negative grid voltage to bring the CRT beam to cut off - solution; add turns around LOPT core to gain more negative bias for the CRT grid. Requires one 2.5mm hole added to the pcb for the wires to pass through.
4) Turn off spot problem- solution increase C22 to 220uF
5) Protect VDU from damage from abnormal H drive signals- solution; add 5W rated zener diode to H output stage.
So to do these things I made a sub-board which carries modifications 1,2 & 5, from 1/8" fiberglass with some silver plated brass eyelets, though it could readily be a standard thickness pcb (if somebody makes a gerber). It mounts above the LOPT, using an extension above one existing pcb mounting hole and requires only one 3mm hole added to the pcb for the other mounting post.
The H linearity, without the modification on the PET VDU, is pretty poor, as expected, stretched on the left side. The technical reasons for this linearity error and why this added diode improves it are explained in the article I'm doing. It is not widely known or published anywhere that this technique of combining a Germanium energy recovery rectifier with a Silicon HOT is a linearity solution. I discovered it myself back in the 1970's, it significantly improves the left sided scan linearity defect , see picture. It is not practical in the PET to improve the H scan linearity with the common method of using the magnetic linearity coil, because there is limited scan width & no width control.
Adrian, from Adrian's digital basement video, about the PET VDU, also noticed the CRT focus was soft. He added a pot in attempt to adjust it, but the voltage required to get the CRT through a focus voltage knee is not present on the pcb, you can only get it between 0 and 85V. These CRT's achieve corner focus at about 420V on G4 and center focus at about 220V. (this difference is due to the extra length the beam travels to get to the corners). However a good compromise is around 330V, but for any individual CRT, it really requires a control pot to allow for CRT variation and to get the perfect compromise between center & peripheral focus (some VDU's have focus modulators to correct this center-corner difference, but adding that for this VDU would be going too far, even for me). Initially I built some voltage multiplier circuits, however the superior solution was to step up and rectify the HOT's collector voltage to around 580V with an auto-transformer wound on a 1" pot core with very easy to handle 30 awg wire and a UF4007 rectifier. The potentiometer was selected to be suitable for the task, high quality Japanese 2M ohm 0.25W rated part.
Also, one thing that helps is to get the CRT out of the cabinet and make a stand for it, I used some 3/8" aluminium bar. This makes life much easier working on the VDU. Also adding single pin connectors helps with the yoke wires and those for the mod board so they can be connected & disconnected at will.
The screen shot shows the VDU working with all the mods in place using X's to make a test pattern. Generally after the mods the performance is pretty good and the individual pixels comprising the X's can be seen clearly at the periphery and at the CRT center.This is an original Amperex MW24-302GH CRT that came in the spare VDU. There is some mild barrel distortion of the raster more in the global upper R corner area, this yoke does not have any overall geometry correcting magnets on it as they were obviously keeping the cost down, hence the absence of a focus potentiometer.
The article is very large and not finished, because of the extensive work on the LOPT analysis (directed toward using substitute LOPT's if needed) so I thought I would post some of the other work more briefly. The idea was to make the modifications reversible with as little change to the pcb as possible.The things that require improving are:
1) Poor Horizontal scan linearity - reason, no magnetic linearity inductor- Solution add high speed high voltage germanium rectifier and alter S correction capacitor value to 8uF.
2) Poor focus- reason, inadequate focus voltage for G4 electrode, needs to be adjustable around 330V. Only GND & 85V available inside VDU - solution; add transformer/rectifier/focus control powered by 580V.
3) Inadequate brightness control range (both VDU's same problem)- reason, the manufacturer's "cheated" with the design of the video output stage because the voltage ratings of the chosen output transistors were not high enough for normal CRT cathode voltages around 70v and also they did not have a higher first anode voltage than 85V (ideally for the CRT ist anode its 120V if the cathode voltage was higher around 70V) so they were forced to regulate the cathode voltage down with a zener diode. This required a higher negative grid voltage to bring the CRT beam to cut off - solution; add turns around LOPT core to gain more negative bias for the CRT grid. Requires one 2.5mm hole added to the pcb for the wires to pass through.
4) Turn off spot problem- solution increase C22 to 220uF
5) Protect VDU from damage from abnormal H drive signals- solution; add 5W rated zener diode to H output stage.
So to do these things I made a sub-board which carries modifications 1,2 & 5, from 1/8" fiberglass with some silver plated brass eyelets, though it could readily be a standard thickness pcb (if somebody makes a gerber). It mounts above the LOPT, using an extension above one existing pcb mounting hole and requires only one 3mm hole added to the pcb for the other mounting post.
The H linearity, without the modification on the PET VDU, is pretty poor, as expected, stretched on the left side. The technical reasons for this linearity error and why this added diode improves it are explained in the article I'm doing. It is not widely known or published anywhere that this technique of combining a Germanium energy recovery rectifier with a Silicon HOT is a linearity solution. I discovered it myself back in the 1970's, it significantly improves the left sided scan linearity defect , see picture. It is not practical in the PET to improve the H scan linearity with the common method of using the magnetic linearity coil, because there is limited scan width & no width control.
Adrian, from Adrian's digital basement video, about the PET VDU, also noticed the CRT focus was soft. He added a pot in attempt to adjust it, but the voltage required to get the CRT through a focus voltage knee is not present on the pcb, you can only get it between 0 and 85V. These CRT's achieve corner focus at about 420V on G4 and center focus at about 220V. (this difference is due to the extra length the beam travels to get to the corners). However a good compromise is around 330V, but for any individual CRT, it really requires a control pot to allow for CRT variation and to get the perfect compromise between center & peripheral focus (some VDU's have focus modulators to correct this center-corner difference, but adding that for this VDU would be going too far, even for me). Initially I built some voltage multiplier circuits, however the superior solution was to step up and rectify the HOT's collector voltage to around 580V with an auto-transformer wound on a 1" pot core with very easy to handle 30 awg wire and a UF4007 rectifier. The potentiometer was selected to be suitable for the task, high quality Japanese 2M ohm 0.25W rated part.
Also, one thing that helps is to get the CRT out of the cabinet and make a stand for it, I used some 3/8" aluminium bar. This makes life much easier working on the VDU. Also adding single pin connectors helps with the yoke wires and those for the mod board so they can be connected & disconnected at will.
The screen shot shows the VDU working with all the mods in place using X's to make a test pattern. Generally after the mods the performance is pretty good and the individual pixels comprising the X's can be seen clearly at the periphery and at the CRT center.This is an original Amperex MW24-302GH CRT that came in the spare VDU. There is some mild barrel distortion of the raster more in the global upper R corner area, this yoke does not have any overall geometry correcting magnets on it as they were obviously keeping the cost down, hence the absence of a focus potentiometer.