HP-35 Power Supply Waveforms

[cruff]

With the talk about the HP-35 being powered from a LiPo battery, I got interested in how the power supply actually worked. So I constructed a similar version from parts so I could probe it with the oscilloscope and understand how it works. I used the schematic from the HHC 2016 HP calculator power supplies presentation as a guide.

I managed to find a suitable toroid that came close to the inductance values of the schematic along with 2N3904 for the NPN and an old RS-2021 for the PNP transistors. Interestingly, I didn't get the circuit to behave and approach the design voltage outputs until I replaced C2 with a 47 nF capacitor in place of the specified 470 pF.

The diagram shows the waveforms captured at (yellow) Q1 base, (light blue) Q2 base, (purple) the "top" tap of L1, (dark blue) the L2/L3 tap. Specifically, I was interested in what drove the oscillation (a blocked oscillator). This is done by the saturation of the magnetic field which causes the voltage at the top tap of L2 to collapse, driving a negative pulse through C2 to the base of Q2, forcing it hard off (diode D1 acts as a reverse voltage clamp to protect Q2 from a large reverse base-emitter voltage). Q1 acts as the bootstrap for the base of Q2, starting the first cycle, as well as tuning the frequency to keep the Vcc voltage (+6) at the desired value.

When my test circuit was driven with 3.6V at 100 mA, the frequency was about 67 kHz, the extremum values of the waveforms were:
Q1 base: 2.457V, 3.250V
Q2 base: -1.187V, 1.25V
L1: 0.156V, 8.125V
L2/L3: 0.156V, 6.25V

And the resulting voltages were: -9.22V, 5.22V, 7.03V.
   

[albertofenini]

Hi Cruff
this is very interesting, we are in the process of rebuilding a HP65 PSU
and we are having some difficulties in finding some replacements, like the toroid
Can you tell us what did you use ?
Also replacing the transistor is proving to be hard.
The 1854-0550 can be replaced with a 2N3904 or a more easy to find BC547 (reverse pin out)
The 1954-0094 however can be replaced with PN3646 which is very hard to get
Do you think we could use the RS-2021 transistors like you did ?
Thank you very much for help !

[teenix]

(03-28-2019 04:06 PM)albertofenini Wrote:  The 1954-0094 however can be replaced with PN3646 which is very hard to get

You can replace the PN3646 with a 2N3703, MPS3638 or PN3638

PN3638 similar to PN200

PN3904 similar to PN100

cheers

Tony

[cruff]

(03-28-2019 04:06 PM)albertofenini Wrote:  Hi Cruff
this is very interesting, we are in the process of rebuilding a HP65 PSU
and we are having some difficulties in finding some replacements, like the toroid
Can you tell us what did you use ?

My version was not meant to be an exact replacement, but just to work similarly so I could capture the wave forms. One transistor is a RS2021 which is apparently equivalent to the NTE159 or 2N3906. The other transistor is a 2N3904. The torroid was something I had in my parts stash, but I don't know the manufacturer nor its specifications. As you can see it is just on a breadboard.

   

[albertofenini]

This what we are doing (actually my son is doing this)
Is the basic block for the oscillator made with spare parts from old PSUs
the aim is to test on field the quality of each component and then transplant them in this
project Rebuilding an HP65 logic board from scratch
Over the weekend we should be able to add the diodes, the capacitors and the second transistor
and see how close we are in generating Vcc, Vgg, Vss from Vbatt
If we will succeed we will add ROMs and we will try to rebuild a CPU which was good but had
some loose pins ...
Will see and keep posting !
Still we have not found a replacemt for the 18154-0094 and in this prototoype an old genuine 1854-0550 has been used ..

   
   

[albertofenini]

Here we go !!
Edoardo has worked very hard and the HP65 PSU has been reproduced on a breaboard !
We used mostly original spare parts, the Toroid, the transistors, the resistors, couple of capacitors are original.
We replaced the 3,3 microF capacitors, with new parts very close with the value (about 3,8 microF)
We replaced the 15microF with a new part very close to that
We replaced the diodes, the original have a value of about 0,287 V while these new have a value of 0,347 or 0,37Volts
(BTW : why the voltage value of a diode displayed on the multimeter is a 1/10 of the real value ??)
We are attaching pictures and a drawing we used to check the correctness of the PSU
Everything seems fine but ....
Vbatt = 3,9V
Vcc (expected 8.2V) is 8V (good)
Vgg (expected -12,4 V) is -37V !!! (three times)
Vss (expected 6.28V) is 15,8 V (more then double)

Any help in finding what we did wrong or what is missing is welcome !!!!
Thanks in advance !!

   
   

[cruff]

Do you have the phasing of the transformer toroid correct? That would easily throw off the voltages.

[albertofenini]

Yes, we think so, and we have double checked that with the original HP65 PCB,
however, we may have mixed up some Toroid coming from Classic's PCB with some
Toroid coming from Woodstock's PCB, are they all equal ?
And if not how can we distinguish the two tipes ?
Also, with do not have and inductance meter, how an we check if the Toroid are good ?
Thanks help !!!

[albertofenini]

We double checked and the scheme seems ok,
BTW, I'm right if I consider that Vgg is wrong because Vss is wrong ?
I understand well that these two are somehow related ?
Reading some of the articles it looks like Vss is a reference Voltage.
When Vss is wrong we have read that Cr4 (the diode cathode connected to the cathode of the zener) is a suspect.
Is this a good theory or we are totally wrong ?
Thanks for help !!!!

P.S. : one last question, are the Toroids from classic equal to the toroids from Woodostock ?
Looking here https://www.keesvandersanden.nl/calculat...supply.php
it seem that L1 has a different value if is from a classic (190H) or from a Woodstock (110H)

[cruff]

I noticed in your other thread that the image of the 65's power supply is cut off on the left side. Just to make sure, you have a connection between Vss at the junction of C7/CR5 and CR4? If that is missing, the voltage regulation of Vss and Vgg would not work. CR 4 is connected in the correct orientation? What is the voltage across it while the supply is running (oscilloscope)?

As for checking the windings of T1, the number of turns for the L2 and L3 windings will be the same (as will their DC resistance). L1 will have more turns and the DC resistance will be a bit higher than for L2/L3. From your picture with the green perfboard, L2/L3 are green and red, L1 is the coppery colored winding.

[cruff]

In addition, if you measure the windings from point 1 to 2, 1 to 3 and 1 to 4 on the schematic picture, in that order, the DC resistance will increase linearly. It isn't clear to me from the pictures I've seen of the toroids and their windings if the interconnections between L1, L2 and L2 and L3 are separable, such that it would be possible to mix things up when placing them into a new circuit.

[albertofenini]

We re-designed the circuit on a different breadboard using the same components.
At the beginning we were having the same values, good Vcc and bad Vgg and Vss.
We decided than to change the zener, although the one used was coming from a donor
and the multimeter was giving it as working.
We tried another one (also from a donor) but the transistors were getting hot ...
We tried a replacement with a measure of 0,7 volt (which still I don't understand why it means a
voltage cut of 7 V) and everything worked fine.
Vbatt is 3,94 Volts
Vgg is - 15,47 Volts
Vss is 7,21 Volts
Vcc is 9,67 Volts
Values are slightly higher than expected, but I think we can consider it a success
We'll try to post the waveforms soon.
As for the Toroids we followed your suggestion, but L1, L2 and L3 all have a resistance of 1,3 Ohm,
and even if we measure L1+L2, or L1+L2+L3 this value doesn't change
We are however planning to buy a LCR meter or a multimeter with LCR functions,
at that point we should be able to measure the inductance.
One last note about the other thread : http://www.hpmuseum.org/forum/thread-12690.html
The pictured board has a connection problem with both the Ground trace and a couple of
interconnected PSU components.
However, now that with this experiment we can select a complete list of PSU components
proven to be working, we may go back to that project and use external wires to replace
the broken traces.
Thanks again to everyone for all the ideas !

   

BTW, what should we change to get on this PSU the exact values for Vcc, Vgg and Vss and
not these which area little higher ?

[cruff]

Higher voltage zeners can not be tested with an ordinary diode tester. You have to put them into a circuit as they are intended to be used and measure the voltage in the reverse biased direction. Otherwise you will just measure the forward bias diode drop.

If you put a 6V zener into the circuit you should probably see voltages closer to the expected values.

[albertofenini]

We did it !
We created a micro circuit, a resistor, the diode and a 9V generator and measured the real value.

We discovered that the zener with which Vss and Vgg where very high had a voltage of a little more of 9V
With the replacement, which has a voltage of 6V we have Vss and Vgg much more close to the correct values (see pictures in precedent post)

Should we try something lower to get the exact Vcc, Vss and Vgg values ?
Thanks for help, this is getting interesting

[cruff]

(04-05-2019 12:34 PM)albertofenini Wrote:  exact Vcc, Vss and Vgg values ?
Thanks for help, this is getting interesting

Good news. How close are the voltages now? Lacking information on how HP specified the tolerances when procuring the zeners, if it is within 5%, then it may be good enough, since this is not a high precision application. You could tell by running the calculator on the voltage and if it behaved then you would be good to go.

[albertofenini]

With a Vbatt of 3,94 Volts this is what we get
Vcc : 9,67
Vss : 7,21
Vgg : -15,47

Much better that the previous, but still too high, I would not power a calc with these.
The Transistors are original (0550 and 0094), the 2,2nF, the 22microF and the 60microF capacitors are originals and so is the Toroid.
Resistors are a mix of new and original, but all of the exact (measured) value
Diodes, Zener and smaller capacitors are new, of very close value to the original ones.
We may try to find a zener with even a lower value or harvest one from a donor and see how it change,
what do you think ?
Thanks and take care !

[teenix]

(04-06-2019 08:59 AM)albertofenini Wrote:  With a Vbatt of 3,94 Volts this is what we get
Vcc : 9,67
Vss : 7,21
Vgg : -15,47

Much better that the previous, but still too high, I would not power a calc with these.
The Transistors are original (0550 and 0094), the 2,2nF, the 22microF and the 60microF capacitors are originals and so is the Toroid.
Resistors are a mix of new and original, but all of the exact (measured) value
Diodes, Zener and smaller capacitors are new, of very close value to the original ones.
We may try to find a zener with even a lower value or harvest one from a donor and see how it change,
what do you think ?
Thanks and take care !

Not sure what value zener diode you are using, but maybe try a 5.6V or 6V or 6.2V, 500mW. They are all cheap to buy.

cheers

Tony

[albertofenini]

ciao Tony,
we have tried several zener all with a voltage ranging from 5,6 to 5,9 volts and the above values are persistent :
Vcc : 9,67
Vss : 7,21
Vgg : -15,47
consider + or - 0,5 V testing the different zener.

at the beginning when the Voltages Vcc, Vss and Vgg were almost the double of the expected values, the zener was about 9 V
take care and if you have any suggestion, please let us know !

[cruff]

(04-06-2019 08:59 AM)albertofenini Wrote:  Diodes, Zener and smaller capacitors are new, of very close value to the original ones.

Did you use a silicon diode for CR5? That would cause a difference in the voltages. From the supply schematic we have been using, HP used germanium diodes in the original circuit, with their correspondingly lower forward voltage drop.

[albertofenini]

Hi
I'm using for all the diodes some germanium diodes marked AA117, these have a forward voltage drop
of 0,328 Volts versus the original that have a forward voltage drop of 0,275 V
I have a few original from hp left, not enough to make a whole PSU, but I can substitute just CR5
Do you think that changing just CR5 would make the difference ?
Thanks for help!