More Fun With Magnavox: The 9300 Series
- Thread starter dcgillespie
- Start date
knockbill
Lunatic Member
Got an old train transformer? 100w works fine... If not I know what you're gettin for christmas!!!didn't have a handy source of 10v,
8764 seems more like it,,, glad is sorted out...
By your first transformer data post, the transformer is coming out to be about a 13K transformer whether using the 4,8, or 16 ohm taps -- assuming the taps are rated with their rated load impedance. You're definitely getting some screwy numbers: The 4 ohm tap should show a turns ratio that is exactly twice that of the 16 ohm tap, which it does not.
I have always found it much easier to drive the primary, as the voltages are much easier to read. I apply the AC line voltage directly to the plate leads of the transformer, measure what the line voltage is, measure the voltage from the full secondary, and then determine impedance from those numbers. It actually involves less voltage than your approach, but just be careful around the line connections!
Dave
I have always found it much easier to drive the primary, as the voltages are much easier to read. I apply the AC line voltage directly to the plate leads of the transformer, measure what the line voltage is, measure the voltage from the full secondary, and then determine impedance from those numbers. It actually involves less voltage than your approach, but just be careful around the line connections!
Dave
that gives me something else entirely.
120.3vac across the primary gives me 5.0v on the 16 ohm tap, 3.4vac output on the 8 ohm tap, or 2.3v on the 4 ohm tap. Assuming an 8 ohm load, that gives me a 4631 ohms, 10015 ohms, or 21885 ohms assuming an 8 ohm load.
at this point I just have no idea what I have here.
Are there any other ways to measure this? Eventually one will give me an answer I like
120.3vac across the primary gives me 5.0v on the 16 ohm tap, 3.4vac output on the 8 ohm tap, or 2.3v on the 4 ohm tap. Assuming an 8 ohm load, that gives me a 4631 ohms, 10015 ohms, or 21885 ohms assuming an 8 ohm load.
at this point I just have no idea what I have here.
Are there any other ways to measure this? Eventually one will give me an answer I like
Gadget -- A couple of things:
1. To determine what the transformer "is", treat each tap for the impedance it is rated for. Therefore, a properly loaded 4 ohm tap gives 10,943 ohms. A properly loaded 8 ohm tap gives 10,015 ohms, while a properly loaded 16 ohm tap gives 9262 ohms.
2. It is always best to use the full secondary in determining what the intended primary impedance is, since it represents the full winding, and a known impedance. Therefore, it will most accurately represent the intended primary impedance.
3. In a well made OPT, the 4 ohm tap will represent EXACTLY twice the turns ratio of a 16 ohm winding, since the 4 ohm tap represents the CT of the winding. And, as OPT winding goes, it is generally easy to attach a 4 ohm tap as it can be placed at the connecting point of 1/2 of the secondary interleave windings. Therefore, the 4 ohm and 16 ohm taps will generally always reflect exactly the same impedance back to the primary when properly loaded. It is always the 8 ohm tap that is the little bugger, as it may need to be placed literally in the middle of a winding, which generally is not done. Therefore, the 8 ohm tap is usually the one that will reflect something slightly off from that of the 4 and 16 ohm taps.
4. Your transformer however is frankly all over the map, leading me to question the ultimate quality of the transformer. Edcor is one of those brands that everybody always "says" is great -- but who has any hard data to back this up? There is absolutely none that I know of. In any event, with 8 ohms being the mainstay of most replacement transformer designs, I think it's fair to say that your transformer is intended to be a 10K primary -- particularly if you are actually using the 8 ohm tap -- and that it is the reason for your power loss.
Dave
1. To determine what the transformer "is", treat each tap for the impedance it is rated for. Therefore, a properly loaded 4 ohm tap gives 10,943 ohms. A properly loaded 8 ohm tap gives 10,015 ohms, while a properly loaded 16 ohm tap gives 9262 ohms.
2. It is always best to use the full secondary in determining what the intended primary impedance is, since it represents the full winding, and a known impedance. Therefore, it will most accurately represent the intended primary impedance.
3. In a well made OPT, the 4 ohm tap will represent EXACTLY twice the turns ratio of a 16 ohm winding, since the 4 ohm tap represents the CT of the winding. And, as OPT winding goes, it is generally easy to attach a 4 ohm tap as it can be placed at the connecting point of 1/2 of the secondary interleave windings. Therefore, the 4 ohm and 16 ohm taps will generally always reflect exactly the same impedance back to the primary when properly loaded. It is always the 8 ohm tap that is the little bugger, as it may need to be placed literally in the middle of a winding, which generally is not done. Therefore, the 8 ohm tap is usually the one that will reflect something slightly off from that of the 4 and 16 ohm taps.
4. Your transformer however is frankly all over the map, leading me to question the ultimate quality of the transformer. Edcor is one of those brands that everybody always "says" is great -- but who has any hard data to back this up? There is absolutely none that I know of. In any event, with 8 ohms being the mainstay of most replacement transformer designs, I think it's fair to say that your transformer is intended to be a 10K primary -- particularly if you are actually using the 8 ohm tap -- and that it is the reason for your power loss.
Dave
Right on then, 10k transformers when they ought to be 8k. That would make sense.
just for S&G, I got out my other cheapie digital meter. 120.8v in gives 3.8v out, or 8084 ohms on the primary from an 8 ohm load on the 8 ohm tap.
I think I need to buy a better meter. This uncertainty crap is getting on my nerves. I've had my eye on one of those fancypants 4 lead HP bench meters for a bit now. I suspect I know what I'm buying myself next.
just for S&G, I got out my other cheapie digital meter. 120.8v in gives 3.8v out, or 8084 ohms on the primary from an 8 ohm load on the 8 ohm tap.
I think I need to buy a better meter. This uncertainty crap is getting on my nerves. I've had my eye on one of those fancypants 4 lead HP bench meters for a bit now. I suspect I know what I'm buying myself next.
Last edited:
Christmas is just around the corner!
Dave
Dave
jazbo8
Well-Known Member
As Dave said, even if it is 10k, it should not make that much a difference in the output power, yet your readings were far off from "spec", have you tried reading the output voltage with a scope? It tends to be more accurate than DMM's...
Don't necessarily trust my x10 probe to have accurate attenuation, and my scope maxes at 5v/div. Can't measure line voltage that way. It also doesn't have handy-dandy direct readout so I'd be into it for counting divisions. I ran over the x10 probe that was in good shape with my chair accidentally about 2 weeks back and it destroyed it. A set of probes is also on my shopping list. Simple fact is I need some better test gear, and running into a situation where the crap I have won't get the job done is an excuse to buy it.
Dave, If someone were inclined to want to keep the circuit as stock as possible was it ever determined which coupling cap values provided the best bass reponse?
Ah, misunderstood. I put the amp back on the shelf for the moment and I'll get back to it later. At this point I'm not trusting the ohm function on that meter either, and without that an accurate voltage measurement still won't tell me anything of value. I'm more motivated to buy a nice bench meter that doesn't suck and revisit this later. I want to re-measure the performance on those Bogen amps too since those numbers didn't quite jive either.
As I was going through the project, I ran into so many other issues with the stock inverter circuit (basic imbalance, frequency imbalance, stability, etc.) that I never really pursued optimizing the original design of that stage. Moving to the floating paraphase configuration was such an easy and positive contribution to the project that once that move was made, any effort on the original inverter circuit was abandoned. However, I think Gordon has determined that as long as a 10:1 ratio is kept between the two couplings caps, the top cap can be moved up to .01 uF, with the bottom cap then becoming .1 uF -- but hopefully, he'll stop by to verify that I'm accurately depicting his thoughts. Such a move would certainly help the LF performance of the original design, but would do nothing to address the HF imbalance. Therefore, I did not pursue this modification, or test the results of such a change.
At some point, I plan de-modify one channel back to the original configuration (using a 200 ohm cathode bias resistor), so that less aggressive modifications can be tested, although the modifications as currently proposed are not hard to implement. In the power supply, simply jumping out the screen dropping resistor (when needed) and reinstalling the 5U4 rectifier tube would return it close enough to the stock configuration for such purposes. The other channel would remain true to the existing modifications without change, since the EFB circuits could care less how many output tubes it is actually supporting. That will then set the amplifier up as a good test bed for future testing of different OPTs.
Dave
At some point, I plan de-modify one channel back to the original configuration (using a 200 ohm cathode bias resistor), so that less aggressive modifications can be tested, although the modifications as currently proposed are not hard to implement. In the power supply, simply jumping out the screen dropping resistor (when needed) and reinstalling the 5U4 rectifier tube would return it close enough to the stock configuration for such purposes. The other channel would remain true to the existing modifications without change, since the EFB circuits could care less how many output tubes it is actually supporting. That will then set the amplifier up as a good test bed for future testing of different OPTs.
Dave
Just a mention, but that same rising response you got out of yours, I got with mine with the 1500 ohm + 390pf cap not installed. I'm sort of wondering if that isn't strictly related to the output transformer. With that network installed, I'm +0.5db at 20khz, without it was +2db.
With the stock circuit and a 0.015 and a 0.1uf and those Edcor transformers, mine was -1db at 18hz. The stock iron won't get that low, so the caps are definitely not going to be hurting any at those values.
With the stock circuit and a 0.015 and a 0.1uf and those Edcor transformers, mine was -1db at 18hz. The stock iron won't get that low, so the caps are definitely not going to be hurting any at those values.
The stock transformers definitely produce a rising response when the loop is closed, necessitating the step network addition with the modified design. While my unit when stock didn't include the 100pF caps that the component table indicated my unit should include, when I installed them, the rise in response was still quite pronounced. The step network is definitely OPT specific, so using that network with your Edcor transformers in the modified design would almost certainly produce a different response curve than that produced with the original transformers.
Dave
Dave
Dave, that's correct- .01 and .1uf seem to be the best working coupling cap values for a stock 93xx amp. Smaller than that, and bass character suffers (loss of gain/feedback effect), and higher than that, doesn't seem to make any real positive difference.
The only other mod I tried on the phase inverter, was to put a pot in the voltage divider pot driving the paraphase section. This seemed to work pretty well, in getting rid of the gross errors due to the imbalance, at least at low frequencies. We just dialed in the pot, to get lowest overall distortion between 100 Hz and 1000 Hz, IIRC (tested at 100 Hz, then at 1000 Hz, setting the pot so that neither point was overly high compared to the other- and IIRC, there was one setting that seemed to satisfy lowest-distortion for both frequencies in the case we tested). This was with an AMP175 6V6 amp, but the same principle would definitely apply here, if someone wanted to try to test an "optimized" stock circuit...
A .015uf cap will give an open-loop rolloff point (given the stock 470K grid resistor) of about 22.5Hz. Given feedback, it makes sense that you'd get a 1dB down point at 18 Hz (the open-loop rolloff is compensated for, by the feedback loop having less open loop gain to work with,and there being less feedback signal, as a result)...
Regards,
Gordon.
The only other mod I tried on the phase inverter, was to put a pot in the voltage divider pot driving the paraphase section. This seemed to work pretty well, in getting rid of the gross errors due to the imbalance, at least at low frequencies. We just dialed in the pot, to get lowest overall distortion between 100 Hz and 1000 Hz, IIRC (tested at 100 Hz, then at 1000 Hz, setting the pot so that neither point was overly high compared to the other- and IIRC, there was one setting that seemed to satisfy lowest-distortion for both frequencies in the case we tested). This was with an AMP175 6V6 amp, but the same principle would definitely apply here, if someone wanted to try to test an "optimized" stock circuit...
A .015uf cap will give an open-loop rolloff point (given the stock 470K grid resistor) of about 22.5Hz. Given feedback, it makes sense that you'd get a 1dB down point at 18 Hz (the open-loop rolloff is compensated for, by the feedback loop having less open loop gain to work with,and there being less feedback signal, as a result)...
Regards,
Gordon.
Last edited:
Morris -- You can't find them because they aren't there. The conditions requiring their addition become significant when the old can cap is replaced with modern caps of the same or greater value, and particularly so at the point where the screen grids are supplied from. Since I was still using the original (good) can cap for development work, and the section of that cap supplying the screen grids became the old original 10 uF section that used to supply the phase inverter stage, the installation of the Screen Stability resistors wasn't required in the development mule at this point -- yet.
The best way to install the resistors is to connect them between pins 8&9 at each output tube socket, and then daisy chain all the pin 8s together and over to the screen grid supply point supplying the screen grids.
You may notice in my unit that I used pin 8 of one of the output tube sockets in the test mule as a tie point for the OPT CT connections. Obviously this pin could not be used for this purpose when the Screen Stability resistors are installed, so an additional tie point would need to be installed to facilitate that connection.
I hope this helps!
Dave
The best way to install the resistors is to connect them between pins 8&9 at each output tube socket, and then daisy chain all the pin 8s together and over to the screen grid supply point supplying the screen grids.
You may notice in my unit that I used pin 8 of one of the output tube sockets in the test mule as a tie point for the OPT CT connections. Obviously this pin could not be used for this purpose when the Screen Stability resistors are installed, so an additional tie point would need to be installed to facilitate that connection.
I hope this helps!
Dave