Re: bad bottlecap transistors causing squiggly vectors?

From: William Boucher <boucher_at_mnsi.net>
Date: Fri Jul 30 2010 - 00:46:37 EDT

I just thought I'd jump in here and talk about the idea of "matching" the deflection amplifier transistors. Don't bother, it won't make any difference. The deflection amplifiers of all vector monitors, that means all models by all manufacturers, are designed as simple voltage-to-current amplifiers. The video signal (X or Y) enters through a simple resistor voltage divider that includes a pot so you can fine tune the overall size for that axis. The input stage where the trimmed video voltage signal goes in is a balanced differential pair of small transistors. The input voltage is passed through this first stage and then converted to a current by the transistor in the second stage. The final stage boosts the current up to the level required by the deflection coils. The final output stage has a current-sensing resistor. As deflection coil current passes through the current-sensing resistor, a small voltage is generated across the resistor relative to ground. This voltage is used as the feedback signal that goes back to the balanced differential input stage (very first stage) to compensate for any error in current gain that might occur throughout the entire amplifier.

Also consider that the deflection coils cause deflection by generating a current that produces a magnetic field. The voltage across the coil is not regulated and you wouldn't want it to be. The current through the coil is regulated using the closed-loop circuit that I described in the above section. This is because beam deflection is proportional to coil current, not voltage. The deflection coils are simple copper wire (don't say duh, I'm making a point :-). The resistance of copper increases as temperature increases. An increase in temperature of approximately 150 degrees C will double the resistance of any strand of copper wire. That said, imagine how much the picture size would shrink as the machine warmed up! Probably as much as 50%. The picture would visibly change size in both axis' as the average room temperature changed. Even varying the distance between the machine and the wall behind it would make a noticeable difference. Hence the use of the constant-current loop to prevent such drift. No vector monitor would work satisfactorily without this feature.

The typical current gain (specified in a datasheet as hfe) of a transistor varies greatly between parts, even those having the exact same part number. The variance can be as much as 100% from part to part. It's not poor manufacturing. It's just a fact of life with respect to bipolar transistors. A specific transistor, say a 2N3904 that has an hfe of 75 is not worse with respect to performance or reliability than another 2N3904 with an hfe of 150. Circuits that are very sensitive to changes in current gain do often require you to use a curve tracer to determine whether or not a specific part will be adequate for your purposes. This was very common practice 40 to 50 years ago when having a single transistor in your circuit was considered a significant achievement. Such requirements are very rare today because mass producing such a circuit would be very time consuming and costly. That is assuming that you want all of your boards to work and perform the same. Such matching of transistors was in the past important in radio circuitry and very simple push-pull amplifiers that lacked any feedback circuitry. These days, circuits are simply not designed in a current-gain dependent way if can be avoided. This has been the case since transistors became more cost effective, say around 1965.

Another big common problem with bipolar transistors is a little thing called "thermal runaway". As the temperature of the transistor increases, it conducts more current. If unchecked, the increase in current will increase the power dissipation thereby generating more heat and the vicious cycle continues until the transistor becomes so hot that it literally becomes incapable of turning off. In most such cases, it soon dies. Fortunately, the closed-loop current amplifier easily compensates for such change in current gain as the transistor warms up. As long as the transistors are not allowed to get too hot under normal operation, they'll be safe. Thermal runaway is also the reason why transistors are never connected in parallel to each other. One will always hog all of the current while the other ones does very little work. This is not the case with voltage-controlled MOSFET transistors which are commonly connected in parallel such as in electric motor drivers or high-power PWM audio amplfiers.

Another worthy note that most of you are already aware of (but that newbies may appreciate) is that the current-sensing resistors on a deflection board (or chassis) are those big white (or beige) cement block low value (i.e. 1.5 ohm) resistors. On the schematic, they are marked "N.I." which means "non-inductive". The inductance of a component causes it to resist a change in the value of the current passing through it. In the case of a deflection amplifier, because the resistor is part of the feedback circuit, this would result in a phase error (lag time) in the feedback signal. This would result in a slight oscillation (jitter) of the current output and you would see this directly manifested as "squiggly vector lines". I have seen a few times where people wrote that a "non-inductive" resistor means that it is "not wire-wound". This is false. There are many makes and models of wire-wound resistors that are non-inductive. This is achieved by constructing the resistor such that every wire coil that goes in one direction is countered by another coil that goes in the opposite direction. In the case of metal film resistors, the resistive metal film is printed in a zig-zag pattern that is printed on a flat ceramic substrate (to make a surface mount part) or a ceramic cylinder (to make an axial leaded part). You have to check the component datasheet to see if it qualifies as non-inductive. In the case of metal film, N.I. is typical and is 99.9% the normal condition. For wire wound, you should always check. Non-inductive components typically cost a bit more because the winding process is slightly more complex.

On a Cinematronics monitor, you'll see two banks of 6 resistors each connected in parallel. These were carbon resistors from the factory. They were carbon because this composition is non-inductive and very cheap. However, carbon has a terrible temperature coefficient meaning that the value of the resistor changes quite a bit with temperature. They won't get very warm during normal operation (that's why there are 6 spread apart and not just 1 fat one) so there won't be much drift in the picture size. There will be some drift however which is just one of the several reasons you should let the machine warm up completely before adjusting anything. Whenever I rebuild a Cine monitor, I replace the carbon resistors with non-inductive metal film components which are extremely temperature stable. This completely eliminates the picture size drift.

If anyone is interested in learning what alternative transistors can be safely used in vector monitors, check out my lists below. I do not sell these parts but the links on my pages will take you directly to the parts on one of the best distributors in North America www.digikey.com . There is also a Canadian site www.digikey.ca . The shipping cost is a flat $8 and delivery is usually the next day, some areas 2 days.

For all Atari B&W or Color vectors (Wells Gardner or Amplifone) and for all Cinematronics B&W vectors:
http://www.biltronix.com/arcade_electcomp_01.html#transistor_subs

For Sega GO8-003. These will work for Atari and Cine but I didn't list the PNP compliment to NPN. It's in the same family so it's easy to determine. You will be amazed how many subs are available that are truly superior to the original parts.
http://www.biltronix.com/Star_Trek_08.html#transistor_subs_2n6259

For Sega GO8-003 to replace MPSU10 and MPSU60. These pin-matching modern superior subs cost about $0.50 each while old NOS originals cost about $4 to $8 each.
http://www.biltronix.com/Star_Trek_08.html#transistor_subs_MPSU10

Game On! :-)

William Boucher
www.biltronix.com
  ----- Original Message -----
  From: Pat Danis
  To: vectorlist@vectorlist.org
  Sent: Thursday, July 29, 2010 2:54 PM
  Subject: Re: VECTOR: bad bottlecap transistors causing squiggly vectors?

  Perhaps I should have been more specific. I don't blindly replace transistors. If the originals are good, I keep them. (Too expensive) If I find a bad one, I do replace them in pairs to make sure the tolerances are the same. I don't profess to be an expert in electronics but think that you would like to have evenly matched parts. I'm sure one vendors tolerances are different from another's. In the description of the wavy lines I referred to, it is hard to determine where the problem lies. The specific case I am referring to is for a Space Duel monitor I rebuilt. The cabinet was a project I purchased as non working. Rebuilt brick, (bad rectifier, fuse block bad and hacked wiring), AR board toasted, and monitor non-working. After rebuilding the AR and brick and fixing the monitor, I could not be sure the if the problem was due to the monitor or the main board or something else. I tried another monitor with a jumper harness and the "waves" were much less obvious but still there. I don't recall the origin of those transistors on the better monitor. I have also found that the replacement transistors seem to be less tolerant of Perhaps it could have even been classified as jitters. I didn't have time to troubleshoot further as I was on a time schedule and had to have it delivered in working in a matter of days. I was lucky to get it done as it was due to the hacks on the monitor.

  I only offered to test a theory I had as a means of contributing to the group. I will still attempt to document my findings and then submit them here for further abuse/discussion. I would also like to point out that I do remove and reinstall the original bottle caps if they are still good but 30+ years of heat, smoke, dirt and dust do take their toll on the heat sink compound and I am doing what I can to keep them running for another 30. Who knows, maybe I'm doing something wrong and I'll find it out. As far as buying in matched pairs, I don't know if the vendor I purchase from has "matched" sets. I only purchase bulk quantities and hope that the tolerances are close enough as they came from the same batch. You're right, I don't have a curve tracer. Wouldn't know how to use one if I had one.

  I still have access to that game and may go by and grab the monitor and test it with a different game.

  I also recently found a monitor out of a bulk buy with transistor numbers I wasn't familiar with. I will dig it out and see what they cross to. I would be interested in seeing what somebody else was using to get a monitor working. For all I know, they are direct crosses. Just wasn't familiar with the part numbers.

  Do you have a source for matched transistors at a "reasonable" price?

  Pat

  Kevin Moore wrote:
    I think Generally speaking it's also a good idea to used matched pairs on your push pulls too. But since buying MP's are not always convenient, and not everyone has a curve tracer to try and match them. I'd have to agree with Mark though. I doubt that would be your problem. They would/could cause other issues. For example, Walling on Tempest.

    Kevin

    On Thu, Jul 29, 2010 at 10:48 AM, Mark Shostak <shostakmark@gmail.com> wrote:

      Uh, kind of obvious, but you could put the original TO-3 transistors back in, and see if the problem goes away. If you're like many of us, the parts are still on the bench...

      However, IMHO it's unlikely that it's the transistors causing the issue, assuming you replaced them with the original devices and not subs.

      Also, rule of thumb in push-pull is to replace _both_ outputs at the same time, and also to replace both drivers when replacing the outputs. This rule primarily pertains to repairs after a failure and not to, "it ain't broke, but I'm fixing it anyway" type activity.

      -Mark

      On Thu, Jul 29, 2010 at 10:37 AM, Pat Danis <patdanis@verizon.net> wrote:

        I too had noticed that some of my WG 6100s were not as sharp after replacing all the transistors and doing a complete rebuild of all the components. The lines were a little wavy and not completely straight. I suspected that perhaps the bottlecap trannies might be an issue but don't have the smarts necessary to make a determination if it was the transistors or something else I might have replaced during the rebuild. I am going to track my findings in some 6100 rebuilds I will be doing in the next few weeks to see if there is a correlation to the transistors I am using. I have 2 different sources for the present time.

        I will see if there is any visible difference between original, and the 2 sets of replacements transistors after a rebuild. My month of August is pretty booked up right now but I need to get some 6100s done so I will attempt to get this info posted. I'll even take video if possible. No guarantees.

        Pat Danis

        PJ wrote:
          I was looking over Darren Finck's post regarding issues with a G05-802 having amongst a number of things, squiggly vectors. This tweaks my memory of the issues I had with an Amplifone that had a bad case of twisted distorted vectors on certain objects (see my starwars posts from last spring). I also had a similar issue on a G05 with original transistors that I figured were getting tired, and once replaced, the issue went away.

          The Amplifone in question had new transistor pairs on both channels. Since the original transistor pairs were long replaced prior to my involvement with whatever transistor the previous owner had on hand (literally), I replaced them with the correct parts, and from evidently, the same batch that I used in the G05.

          What I ended up with was when the drive from the game board was normal, and the gains on the deflection were set higher to fill the screen, the vectors were distorted, like the start and stop rates were bad. Tweaking the drive from the game board to a high level, and dropping the gains on the deflection board made the effect less evident.

          I'm wondering if the 2N3716s and 2N3792s have changed or the specifications aren't what they once were. Anyone else seen this?
        --------------------------------------------------------------------------- ** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org ** Please direct other questions, comments, or problems to chris_at_westnet.com

  --------------------------------------------------------------------------- ** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org ** Please direct other questions, comments, or problems to chris_at_westnet.com

---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
Received on Fri Jul 30 00:45:47 2010

This archive was generated by hypermail 2.1.8 : Fri Jul 30 2010 - 10:50:01 EDT