DRO/Electrical Question

[Bob Brown 1Participant @bobbrown1]

Quote: taken from Shumatech web site

"WARNING!!! The DRO-350 is an old design that uses obsolete parts. The DRO-350 has a number of problems including scale jitter, rounding errors, and axes that won't display. These problems will not be fixed and all future development activity will concentrate on the DRO-375 and DRO-550. The new DRO's completely solve all of these problems and adds a huge list of new features."

[HasBeanParticipant @hasbean]

Hi Les,

The 2032 battery was out of the scale and the power was being supplied by the remote display. On pulling out the cable from the display the scale works for approximately 8 secs before powering off.

I've got some 1F electrolytics on order so well see what happens when they arrive.

Regards,

Paul

[Les Jones 1Participant @lesjones1]

Hi Bob,
Scott has been trying to kill off the DRO350 since the time he developed the DRO550. There have been more problems sourcing parts for the 550 than the 350. It is true now it is difficult to obtain the LED driver chips for the 350. Since the cause of the scale jitter was identified around May 2008 there are at lest three ways to modify the ground path which clears this problem. The non displaying Z axis problem I told Scott what was the cause ant suggested a couple of ways to fix it which involved just a few lines of code to clear flag bits. The rounding errors I have fixed on my own DRO350s. About the time the DRO550 was being released Scott initially allowed me to put a version of code running on a PIC18F2525 (Which just plugs in in place of the PIC16F876) which combined the lathe and mill versions in the file section of the Shumatech forum. He then removed it and asked me not to distribute modified code. I recently asked on the forum if it was OK to distribute modified code now that the DRO350 was a dead product (In his words. ) He blocked the post and did not even answer the question.

Les.

[Les Jones 1Participant @lesjones1]

Hi Paul,
The scale will be working for the 8 seconds due to stored charge on capacitors within the scale. Putting larger capacitors across the power rail in the scale will just make it work for longer than 8 seconds after removing the power.

Les.

[I.M. OUTAHEREParticipant @i-m-outahere]

So we are looking at a voltage drop issue then ?

Paul,

The main issue a was having was with the display flickering and the readout was very unstable which indicated to me there was an electrical noise issue .

If your display is stable the noise issue may not be present with your unit and the 0.02 offset seems to only be present when the battery is removed from the scale so maybe the scale is not getting the correct voltage from the display which could be something simple like a bad connection in the cable or maybe the manufacturer has not accounted for internal resistance etc , maybe you could plug the scale into the display and take a voltage measurement at the battery terminals on the scale ( with the battery for the scale removed ) to see what you get .

Ian

[Ian PParticipant @ianp]

Les

Your understanding and explanation of the 0.2" error is probably correct. I have one caliper that I use which I have removed the plastic overlay from. I can see all the PCB fingers but never thought to measure the spacing/pitch!

As to how it reads, counts and stores the passing of the 0.2 increments is a mystery to me but I happily accept that it works.

I now wonder if all calipers of this type use the same 0.2 pitch, or whether it varies between manufacturers (thats not a question, just me musing.

Ian P

[Douglas JohnstonParticipant @douglasjohnston98463]

I built a Shumatech 350 a number of years ago when the cost of commercial units was quite high. I soon wished I had'nt bothered since the system never worked very well and the price of commercial units came down considerably.

After a few years I bought a commercial unit with glass and magnetic scales and there was simply no comparison with the Shumatech system. The new system is rock steady and has performed perfectly for a number of years.

I know price has to be considered but if funds are available I would urge others to go down the commercial/glass/magnetic scale route and you won't regret it.

Doug

[Les Jones 1Participant @lesjones1]

Hi Ian (P),
As far as I know all of the Chinese scales use the 0.2" spacing of the fingers on the scale bar. What is different is the communications protocol from the data port. I understand the the system was invented by a Swiss or Swedish company and the Chinese copied it. The system was originally called the "Sylvac Inductive System" Here is a link to a description of how it works. (I have not yet read it.)

Les.

[HasBeanParticipant @hasbean]

Right Gents, just an update.

Firstly I'd like to thank you all for the info and advice, it was very much appreciated.

I've finally managed to get the display almost stable. It now flickers occasionally but only by 0.01mm, which I can certainly live with, as there's no way I can machine to that accuracy in the first place!

The fix, however, you can pick and choose from the following items across the scale battery terminals;

1000mF electrolytic cap, 1F supercapacitor or a rechargeable 2032 cell. Any one of these stabilises the display (almost).

I just wish I knew why it was doing it and why fitting the normal CR2032 cell it 'fixes' it

Regards,

Paul

[Ian PParticipant @ianp]

Paul

I've gone back and read your original posting and am still unsure of the setup you are working with..

I get the impression that you have two devices, a scale, and a display. If this is the iGaging type of kit it would only have one power source (in the display) but you seem to have batteries in both.

Were they intended to be compatible with each other? if so the manufacturers documentation should explain the requirements.

How far apart are the two devices, and is the cable supplied for the purpose, or one you have made up?

The only thing I can suggest is that you remove the gear from the lathe/mill and test it in an electrically 'quiet' environment. The jitter you have might be caused by interference picked up from a switching PSU or a VFD.

Ian P

[Ian S CParticipant @iansc]

You might get rid of, or reduce the jitter by looping the cable through a ferrite torroid (half my computer cables are done that way). Ian S C

[Neil WyattModerator @neilwyatt]

Made a few tweaks to my readout last night and this morning. The problems I'd been having were unreliable startup and occasional freeze up or large 'jumps'.

I added 100u caps inside the battery compartment of each scale. In my case the main cause of glitches is switching the mill on and off; multiple starts now failed to make a glitch.

I fitted 470n caps at the readout end of the scale supply wires – I couldn't bear removing the scales again, and they don't seem to be essential, at least not in my case. I think the spikes come through the PSU anyway, so stop them before they run down a cable with the data wires, not at the far end.

I had an LED/diode regulator for each scale, which all gave a good steady 1.67V (I had already put in the dropper diode when I built the units, it turned out). Even so, at Les' suggestion I used an LM317 to a make a separate regulated supply dropped from the 5V rail. Of course, the scales actually need -1.5V dropped down from the 5V rail so I had to recalculate my voltage divider to get 3.5V…

First stab gave me 1.33V which worked two scales, but not the x-axis. Trying to trim to a good 1.5V this I ended up about to chain three resistors and realised a single pot would be easier, and this allowed me to get it to 1.51V

In the end, most of the problems seem to have been caused by a big spike at switch on from the power supply, even though it is a very high-quality regulated one (the transformer & regulator wall warts supplied with Iomega Zip Drives). A 470nF and 220uF cap across where the supply enters the display unit sorted this.

Thinking about the detail, I think there are three likely sources of 'glitch':

Spikes or interruptions to the power rail that upset the scales processor. the 100u cap at the scale itself helps smooth the supply as close as possible to the unit.

HF spikes on the power lines that cause cross-talk with the data lines and upset the readings. the slow value disc cap at the readout end of the cable may help with this.

Power excursions at PSU switch on. Just add extra smoothing where the supply enters the unit.

A few final points:

1 For the record, I dissected my U/S scale head, and there is a direct connection between the supply wires and the battery terminals.

2 The replacement reading head (from Machine DRO) has different behaviour when you use the mode buttions. instead of switching to fast more, it switches to 'hold', pressing the zero button then toggles between 'hold' and 'inc' (incremental?) which is a fast read mode. Press mode again to return to normal.

3 I wonder if these scales are both fast and reliable enough to be used for servo control, whether by stepper motor or ordinary motor?

Neil

[Les Jones 1Participant @lesjones1]

Hi Neil,
I assume the display on the scale is locking up as well as your remote display. When I was about half way down your post I was thinking of suggesting that the scales had somehow got into hold mode but I found that you had already thought of that. The mode can be changed by pulsing the data and clock lines to +1.5 volts. Here is a link to a chart that shows how to move between states, This is quite old so it is possible that some newer scales behave differently. It might be worth disconnecting the data and clock lines at the scale end but still powering them externally to see if the lock up is caused by noise on the power or on the clock/data lines. I seem to remember your code worked in the same way as most others. (Although I think you used triple sampling which I have only seen before in the DRO350 code.) I have only ever used single samples and have never had problems. I did not notice any errors in your code that could cause problems. On the subject of using the scales as part of a closed loop system I seem to remember a discussion on the subject on some other forum (I don't know which one.) and most posts thought it would be difficult to make the system stable due to time delays in the loop. It was also thought that backlash in the leadscrews would cause problems. I have not heard of anyone trying this system so it might be worth a try. It would not be the first time that something was made to work that had theoretical reasons why it might not.

Les.

[Neil WyattModerator @neilwyatt]

Hi Les,

My system has four buttons for each channel.

1 switches between inches and mm

2 goes through display options (which direction is positive, diameter mode, brightness, fractional display etc.) Cycling this also saves the current choice of ins or mm as default.

3 pulls data high

4 pulls clock high.

The Yadro site is where I originally got information on the protocol I could understand. Oddly, it doesn't mention that pulling clock high resets the read head to zero.

I think any servo system would have to have compensation built in to allow for the motor inertia, even with steppers

Neil

[Author]

Posts