# Thread: Broken Cromemco 8k Bytesaver

1. Did he look at the diagram on:

http://www.s100computers.com/Hardwar...001%201978.pdf

for the transformer?
Dwight

2. The regulation is slightly different on this board version but it is the same transformer, XT8K. The pins look to be numbered differently but they should still be wired the same. Your schematic shows pins 2 and 3 numbers swapped ( possible schematic error in one or the other ). Using your schematic, the primary would be 2 to 3 and 3 to 4. The secondary would be 2 to 5. Otherwise it is the same transformer.
Dwight
Last edited by Dwight Elvey; October 29th, 2020 at 01:51 PM.

3. Dwight, for my diagram I went by the pin numbering on the attached diagram.

pin 2 supply, pin 1 feedback, pin 4 collector,, and 3 & 5 the secondary. The circuit you posted on the Bytesaver II has different pin numbers, but the same transformer

It is customary in these sorts of step up supplies to make use of the voltage on the collector, so the secondary is connected there to improve the efficiency of the transformer and requires less secondary turns.

One thing I could have mentioned about this design, unlike some flyback supply designs, the transformer is not run to near magnetic saturation. By applying more base current, the transistor stays on longer for its part of the cycle, the current in the inductor is higher at the moment the transistor switches off, so the stored magnetic energy is greater and hence the magnitude of the voltage pulse on the transistor's collector and secondary is higher. So its easy to regulate the output voltage by controlling Q10's base current via the 1k resistor.

I would also "guess" for a 30V regulated supply the peak voltage should (in the unregulated maximum sate) rise to at least 50V. If the collector peaked to 25V, then its possible the winding ratios could be 1f:1c:1s (f = feedback winding pin 1 to 2, c = collector winding pin 2 to 4, s = secondary winding pin 3 & 5).In other words all winding the same turns numbers. But it could also be something like 1f:2c:3s, if the peak collector voltage was lower, it required a bigger step up and if the feedback winding had less turns than the collector winding to help reduce the chance of zenering the transistor's B-E.

If somebody had a working board it would be easy to put the scope on it to determine the ratios.
Last edited by Hugo Holden; October 29th, 2020 at 03:17 PM.

4. Also.... they fouled up more than one thing on the pin numbers on the Bytesaver II schematic. The circuit there as it stands cannot work because it will not oscillate, there is no 180 deg phase shift on the tap feeding the 680 pF cap and transistors base, so at the bare minimum they got the pins 1 & 2 mixed up.

5. I am trying to LT SPICE model the Royer Oscillator, but have not gotten it to resonate yet.

Using a pulse generator to control Q10 and can adjust the transformer turns ratio to get 30V output from the supply.

6. Originally Posted by Hugo Holden
Also.... they fouled up more than one thing on the pin numbers on the Bytesaver II schematic. The circuit there as it stands cannot work because it will not oscillate, there is no 180 deg phase shift on the tap feeding the 680 pF cap and transistors base, so at the bare minimum they got the pins 1 & 2 mixed up.
The diagram shows no polarity dot of the windings. The secondary can be ignored as it isn't need for the oscillation. I'd need to think about the winding polarities a while but I believe you are right, the polarity of the windings look to be wrong between the rail and the capacitor. The older schematic shows a different regulation but that can be ignored for now.

7. Originally Posted by m_thompson
I am trying to LT SPICE model the Royer Oscillator, but have not gotten it to resonate yet.

Using a pulse generator to control Q10 and can adjust the transformer turns ratio to get 30V output from the supply.
If your spice model doesn't include noise, it may never break into oscillation. You may need to give it an initial kick to start the oscillation. It needs to be enough to get the transistor into the non-linear actions. Try reversing the polarity of the small winding from the rail to the capacitor feedback.
In any case, start the simulation with an initial value other than zero.
Dwight

8. Originally Posted by Dwight Elvey
The diagram shows no polarity dot of the windings. The secondary can be ignored as it isn't need for the oscillation. I'd need to think about the winding polarities a while but I believe you are right, the polarity of the windings look to be wrong between the rail and the capacitor. The older schematic shows a different regulation but that can be ignored for now.
It would not help, the Dot markings, in this case, because it, pin 1, is drawn as a tap on the primary. This implies the polarities of the two halves of the primary without dots. For dot markings showing the winding phases, to be of any use, then the two windings would have to be drawn separately, each having two connections. So clearly this is a schematic drawing an error, where, regardless of pin numbering, the power supply connection should have been connected to the tap on the primary winding.

9. Originally Posted by Dwight Elvey
If your spice model doesn't include noise, it may never break into oscillation. You may need to give it an initial kick to start the oscillation. It needs to be enough to get the transistor into the non-linear actions. Try reversing the polarity of the small winding from the rail to the capacitor feedback.
In any case, start the simulation with an initial value other than zero.
Dwight
Yes this is a real thing with symmetrical circuits like a multivibrator as there is no initial imbalance to start it in Spice. However, in this particular circuit (which is like an asymmetrical single ended mutivibrator) it will always start if there is zero initial charge on the 680pF capacitor and zero initial inductor current.

10. Originally Posted by m_thompson
I am trying to LT SPICE model the Royer Oscillator, but have not gotten it to resonate yet.

Using a pulse generator to control Q10 and can adjust the transformer turns ratio to get 30V output from the supply.

The tricky part with the Spice model is specifying the initial inductance and the winding self capacitances. I'm guessing in a small inductor like this they are unlikely over 50pF. The inductance and the capacity set the self resonant frequency and the peak voltage, along with the energy stored per cycle in the inductor, and the inductor must be large enough that it does approach saturation during the time that the transistor is turned on, we could roughly guess that from the circuit.

One way to start is to choose an operating frequency, say around about 50kHz. A half period is 10uS. Pick an initial inductor , primary inductance say 5mH, the initial current rise will be 5V/5mH =1000 A/Sec. After 10uS the inductor current would be 10mA.

The inductor stored energy per cycle would be ((0.01)^2 x 5mH)2 = 0.25 uJ.

During the inductor ringing cycle the energy is transferred to the self capacitance, so the voltage on the primary (using CV^2/2) at its peak will be about root ( 2 x 0.25E-6 / 50 pF ) or root 10000 = 100v peak (in practice it will be less due to losses).

So I'd recommend in the Spice model, initially make the primary inductance 5mH and its self capacitance in the range of 50pF to get the "model" going. You could distribute the self capacitance over the secondary as well.

If the oscillations don't start, try putting in a value of initial inductor current or some initial voltage (charge) on the 680pF cap and kick start it that way.

I could point out that some styles of circuits with inductors, can be more difficult than others to model in Spice. You could always try an experimental lash up if you don't succeed with the model. Most of the required parameters can be worked out on paper.
Last edited by Hugo Holden; October 30th, 2020 at 03:12 PM.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•