Woodstock power supply

[Paul Dale]

Looking at the woodstock schematic:



Would it be possible to discard the transformer that comes with these calculators and use the cable and plug to provide a rectified 3v or 3.1v supply instead? I'd like to use mains power for a 20-series device without risking burning it out completely. Is a voltage drop on the supply sufficient? Would I also need current limiting?

My, albeit limited, understanding of circuits indicates that it needs 2.5v at the battery and there is a single diode drop along the way (i.e. -0.6 to -0.7 v drop). So 3.1v roughly.

Am I on track here or way off the mark?


Pauli

[AndiGer]

Maybe this thread will help you a little?
http://www.hpmuseum.org/forum/thread-6818.html
Andi

[PANAMATIK]

(01-06-2017 09:23 AM)Paul Dale Wrote:  Am I on track here or way off the mark?

Perfectly on track!

The average charging voltage behind the diode with the original charger is 3.3 Volt when the calculator is switched off and about 3 Volt if the calculator is switched on. Thus a regulated DC voltage between 3.0 - 3.3V (+ 0.6V diode drop) will work and will not cause damage to the circuits when batteries are removed.

if batteries are removed the original charger applied an average of 6.8V to the power supply circuit. This caused damage to the chips.

Keep in mind, that these voltages can potentially cause overloading of the batteries, if applied too long.

see Jaques Laportes video

Bernhard

[hibiki]

I would not discard your power transformer, maybe it can be used in a power supply of 3Vdc:



Depending on the current consumption, a small heatsink attached to LM317T is enough. If your soldering skills are good, an assembly without a PCB can be possible and all components put inside the charger case.

Try and test the power supply circuit in a breadboard, before packing them inside the charger case.

[Paul Dale]

Thanks for the information.

I've got enough information to plod ahead powering recently a refurbished 21 from a non-battery source.

My soldering probably isn't up to putting the regulator into the transformer. I should be up to the capacitors and diodes in the battery box compartment.


Pauli

[jebem]

The HP-21 HP-25 power supply design uses a form of shunt regulation when the external power charger/adapter is in use and the calculator is turned ON.
The battery is the shunt regulator element and it is in parallel with the external power charger and with the calculator DC-DC converter.
The current limiter R1 (8R2) is only inserted in series with the battery when the calculator is tuned OFF and is used to protect the battery.

Shunt regulators are very inefficient, as it must consumes current in parallel with the load to do the regulation and they require a series current limiter of some sort. The load here it is the DC-DC converter.

In this specific basic HP design it requires a current limiter when the calculator is turned ON, otherwise the difference between the transformer secondary voltage and the battery nominal 2.4VDC would simply either destroy the power adapter transformer or kill the battery.

The solution HP found to limit the current in the above conditions was to design the transformer with a very poor regulation by setting a very high secondary winding impedance to drop the excess voltage above the nominal battery voltage.
The dropped voltage is dissipated as heat inside the transformer.

The calculator DC-DC converter alone will consume around 110mA idle and 150mA with all LEDs lit.

So basically the external transformer is under "overload" all the time we use it with the calculator tuned ON.

The HP power charger can deliver a maximum current of 150mA (1.8W at 10VAC under no load, but just 7VAC under a 47Ohm load for 150mA).

So, to make a long story short, replacing the original battery with a capacitor in parallel with a series of 3 or 4 rectifier diodes, will work but I would expect and additional transformer overload that may result in shorter life cycle.
Additionally, because now we have a capacitor instead of a battery, we have an extra ripple current charging the capacitor at each positive half cycle that will also increase the transformer overload. The larger the capacitor capacitance, the larger the ripple current.

That said, I used that modification many times along the years in different types of equipment because it is a easy and fast fix in many situations.

But I admit it is not my first design choice for two reasons:

- Power inefficiency, as explained above.
There is nothing we can easily do to mitigate this.

- Catastrophic failure of one of the rectifier diodes.
Because these diodes are under heavy current load, they can fail by opening the internal PN junction, resulting in a voltage supply overload that can kill the calculator electronics.
We must mitigate this risk by using diodes with enough dissipation power to cope with the current at all times. I would use 1 Amp 100V Si diodes as a absolute minimum, like the 1N4001.

When possible, I would use series regulators as they can be much more efficient and safer but this would either require internal calculator or power adapter modification.

[PANAMATIK]

(01-07-2017 05:39 PM)jebem Wrote:  There is nothing we can easily do to mitigate this.

Not easily, but Yes, we can!

I herewith announce a new circuit for all woodstock calculators, which is easy to install and will consume only 10-40mA when display is on and draw only microamps in standby mode. It will shutdown the original power supply completely and is resistant against overvoltage too. And adds a lot of memory and much more. Schematic is already in its final stage. Should be available end of spring, if all goes well.

Don't give away your woodstock calculators, they will get a rebirth as real power LED machines soon.

Bernhard

[Hans-Peter]

Instead of the LM317 I would take a DC-DC converter that costs about 1$ (seach on e..y):

DC-DC Step Down 4.75V-23V to 1V-17V

The module has a size of 17x11mm and it makes no heat like the 317er.


Hans-Peter

[Paul Dale]

I've ordered one of the DC-DC converters.

Now I'm in a bind, wait for Bernhard's new power supply, wire up a DC supply, modify the AC adapter or even create a USB power cable.


Pauli

[Paul Dale]

WP firmware for a Woodstock LED calculator would be much easier than the 43S. It really would be a case of trimming back the 34S firmware and reworking the display/keyboard code.

Pauli

[PANAMATIK]

(01-08-2017 04:14 AM)emersone12 Wrote:  .. It would be welcome to see the commercial re-introduction of at least one LED model like the HP-25 from the 1970's alongside the new LCD DM-42L.

I agree. Unfortunately this would not be easy. Even with 40mA you need two AA batteries inside or Li-Ion flat accus. It needs the design of a new housing.

After QDSP6064 NOS are sold out, LED Bubble displays are no longer available, only some bigger quite expensive alphanumeric modules I have found and 12 digits would not fit into the form factor of a woodstock case. Not to mention the keyboard which had to be of equivalent quality to the original.

In size and quality the "Woodstocks" were the ultimate LED pocket calculators ever made.

Bernhard

[PANAMATIK]

(01-08-2017 04:51 AM)Paul Dale Wrote:  WP firmware for a Woodstock LED calculator would be much easier than the 43S. It really would be a case of trimming back the 34S firmware and reworking the display/keyboard code.

Pauli

How much memory would be needed for this approach? Which ARM processor would you prefer for the LED 34S?

Bernhard

[Paul Dale]

(01-08-2017 08:45 AM)PANAMATIK Wrote:  How much memory would be needed for this approach? Which ARM processor would you prefer for the LED 34S?

I'll need to discuss this possibility with my partners in crime. The keyboard layout and display wouldn't be easy to cut back but a catalogue based system ought to be possible to increase the number of functions well beyond the 34C (but not the 34S).

The current hardware has 2kb of non-volatile RAM, 4kb of volatile RAM (lost between keystrokes), 128kb of flash and (I think) 55 bytes of display RAM which we use as semi-volatile (it stays alive during powerdown between keystrokes but not a soft shutdown). I think the CPU is an Atmel AT91SAM7L128. All types of memory are stretched to their limits, so some headroom would be very much appreciated. We don't much of the available hardware in the CPU even though quite a bit is run to headers on the 20b/30b hardware. Marcus knows a lot more about this than I do, he is primarily responsible for the hardware support and features.

I don't believe we're tied to this CPU or even an ARM processor. There is essentially no assembly code in the firmware. We do need a C compiler (probably based on gcc) that supports long long ints both for integer mode and elsewhere, we don't need native floating point support.


Is there a form factor suitable 16 segment LED display? We rely on text messages a lot and this would simplify things I suspect -- without this we can do all letters except K, V and X on a 7 segment display if we've got direct access but some important symbols would still be missing (division, multiplication and addition are obvious ones). Going through the anode and cathode driver chips would limit things further but might be possible.

Is a GPS possible? Some SPI FRAM or equivalent for the library? IR printing would be nice of course.


Marcus is very busy with real life, so we'd likely require some assistance on the hardware side. The main firmware shouldn't be a big problem, it will be cutting features and reworking the UI code which I'm fairly familiar with. Which leaves the layout and UI to design.


Pauli

[PANAMATIK]

(01-08-2017 09:51 AM)Paul Dale Wrote:  Is there a form factor suitable 16 segment LED display? We rely on text messages a lot and this would simplify things I suspect -- without this we can do all letters except K, V and X on a 7 segment display if we've got direct access but some important symbols would still be missing (division, multiplication and addition are obvious ones).

I don't see any LED display other than the original. This would limit to 12 7-segment digits. But K,V and X can be displayed with some fantasy. V distinguished from U, X from H, M from N by using small and capital letters.

Code:


k    V    U    W   X    H    M    N 
 _                           _
|_       | |  |_| |_|  |_   | |   _
| |  |_| |_|  |_| | |  | |  | |  | |

(01-08-2017 09:51 AM)Paul Dale Wrote:  Is a GPS possible? Some SPI FRAM or equivalent for the library? IR printing would be nice of course.

Yes, no problem!

[Paul Dale]

M and W are better done using two digits:

Code:


 _   _
| | | | |     |
|     | |_| |_|

X and k are horrible v passable.


Pauli

[Harald]

May I suggest a classic series WP34s?
For the current version I am using a Renesas µPD70F3571, but I'd be happy to port the display driver to a different processor. We could either stick with the 30Bs Atmel (if tha does have a DMA) or go for an ST ARM7.

This would provide us with 35 keys and a 15 digit Display.
http://www.hpmuseum.org/forum/thread-6768.html

[Paul Dale]

(01-09-2017 12:24 PM)Harald Wrote:  May I suggest a classic series WP34s?

Are broken classics commoner than broken Woodstocks? Case and keyboard availability are a concern.

Ideally, 3D printed keys would be used which would be easier on a classic than a Woodstock. The extra row of keys would be a bonus as would the extra display digits but I doubt either is show stopper. The seven segment display is going to be difficult.

At the moment we're thinking about possible layout and interface -- it will take a while to even decide if a LED project is viable. We're not interested in recreating the HP 34C and that pushed the limits of the hardware -- three shift keys, lots of functions.

If it ever does exist, it won't be a WP 34S.


Pauli

[Harald]

(01-10-2017 02:48 AM)Paul Dale Wrote:  Are broken classics commoner than broken Woodstocks? Case and keyboard availability are a concern.

Probably not.

(01-10-2017 02:48 AM)Paul Dale Wrote:  Ideally, 3D printed keys would be used which would be easier on a classic than a Woodstock. The extra row of keys would be a bonus as would the extra display digits but I doubt either is show stopper. The seven segment display is going to be difficult.

I have never seen dual colour 3d printed keys. But if that was possible, it would avoid stick on labels. Other than that I can't see an advantage over using the original ones.
The display indeed is quite limiting. There are several letters that don't work.

(01-10-2017 02:48 AM)Paul Dale Wrote:  At the moment we're thinking about possible layout and interface -- it will take a while to even decide if a LED project is viable. We're not interested in recreating the HP 34C and that pushed the limits of the hardware -- three shift keys, lots of functions.

If it ever does exist, it won't be a WP 34S.

Of course it won't be a 34S, but maybe it could add a lot of functionality over the 34C. For me complex number support is very important.