Loading ...
Sorry, an error occurred while loading the content.

IV-17 / IV-4 Smartsockets

Expand Messages
  • Chris
    As a result of a few enquiries and to suit my own needs I ve been messing about trying to get a smartsocket chip to drive these multisegment tubes. It seems
    Message 1 of 12 , Jan 11, 2009
    • 0 Attachment
      As a result of a few enquiries and to suit my own needs I've been
      messing about trying to get a smartsocket chip to drive these
      multisegment tubes.

      It seems that we can have two aspects from good, cheap, easy, but not
      all three together. Such is life.

      At the moment the 16F690 controller IC is driving a MAX6921 high
      voltage output shift register.

      Of course the problem with shoft registers is their time delay, and
      the IV-17/IV-4 are very responsive, which means getting the last bit
      of dimming when crossfading digits it's impossible to achive and there
      is a noticeable step in brightness between the off state and the first
      level of brightness when a segment is being turned on.

      To improve on that, I clocked the pic from an external 24MHz crystal,
      to get the SPI bus up to 6MHz, and now the fading is very smooth. The
      problem here is that the cost of the crystal could be more than you
      might pay just for one tube !

      An alternative method is to use a parallel driver IC. I've looked at a
      couple of 8 and 10 bit devices which could be driven directly from the
      micro if the software is migrated up to work on a 28 pin PIC. The
      problem I am having here is finding a suitable (not shift regiatser
      based) high side driver which comes in surface mount. If anyone knows
      of one and the cost to drive a whole IV-17 isn't too high, please let
      me know. At the moment I am thinking in terms of using two 8 bit
      drivers for the segments and two discrete transistors for the decimal
      points. A benefit to using this techniques is that two tubes could be
      driven from one controller in a multiplexed style. This reduces the
      cost of components from 1 controlling PIC per tube down to 1 PIC per
      pair of tubes.

      Another technique, although not great, is to use the 'wasted power'
      method of driving. When all segments are lit their total current
      consumption is between 3mA-5mA. What I am thinking of is uysing a
      current limiting resistor for each segment and then a discrete NPN
      pulldown transistor for each segment. This effectively just shorts the
      drive current for segments to ground, via the limiting resistor. I
      would normally not consider this sort of drive system due to it's
      wastefulness, but when talking about such small amounts of waste is it
      still viable ? This would probably be the cheapest method, but is more
      time consuming in the assmebly stage but again it would mean that one
      controller could drive two tubes.

      I would appreciate any comments from the group's members regarding any
      of the above before I start down the wrong path. What would yoiu
      prefer in terms of a suitable design ? Would you rather not have
      multiplexed tubes ? Would you not be bothered if the fading transition
      routines were dropped, or do you like them as they are ?


      Here's the options again....

      a) One PIC16F687/16F690 per tube driving one MAX6921. (requires crystal)
      Makes a small single tube PCB, is relatively quick to assemble.
      Probably most expensive scheme and must have the crystal if good
      quality fading is required


      b) One 28-pin PIC driving two high side parallel to parallel driver IC's
      Slightly more difficult to assemble than a), difficult to get suitable
      chips in SM packages. Can drive two tubes per PCB for one PICand will
      retain all the functionality of the existing Smartsocket. Cheaper than a)

      c) One 28-pin PIC driving 18 surface mount NPN transistors (+1
      resistor each).
      Has all the functional benefits of b) but is more difficult to
      construct due to the large number of small SM components. Probably the
      cheapest option to produce, drives two tubes from one controller PIC.

      Thanks for any input,
      Chris
    • guus.assmann@wolmail.nl
      Hello Chris, For method c you can also use 2 times the 75468 (7 transistors per chip, also in SMD. And additional transistors for the rest. (or 3 times IC,
      Message 2 of 12 , Jan 11, 2009
      • 0 Attachment
        Hello Chris,

        For method c you can also use 2 times the 75468 (7 transistors per chip,
        also in SMD. And additional transistors for the rest. (or 3 times IC, giving
        21 outputs).

        But there's also a method d.
        Use opto-couplers as the level shifters. There's 4 in one housing types.
        And the transistors are capable of 20mA at 70V.
        For routing this is also simple.
        And multiplexing to get 2 tubes with one controller is also an option.

        Fading effects are very nice. If available, they should look perfect.

        BR/
        Guus

        >-- Oorspronkelijk bericht --
        >To: smartsockets@yahoogroups.com
        >From: "Chris" <fixitsan@...>
        >Date: Sun, 11 Jan 2009 10:34:26 -0000
        >Subject: [smartsockets] IV-17 / IV-4 Smartsockets
        >Reply-To: smartsockets@yahoogroups.com
        >
        >
        >As a result of a few enquiries and to suit my own needs I've been
        >Thanks for any input,
        >Chris
        >
      • Quixotic Nixotic
        ... Chris, Farnells have part 1611790, a 24.576MHz surface mount component at £0.192. 10 for £0.147. Farnells have part 1611717, a 24.576MHz a through hole
        Message 3 of 12 , Jan 11, 2009
        • 0 Attachment
          On 11 Jan 2009, at 10:34, Chris wrote:
          > To improve on that, I clocked the pic from an external 24MHz crystal,
          > to get the SPI bus up to 6MHz, and now the fading is very smooth. The
          > problem here is that the cost of the crystal could be more than you
          > might pay just for one tube !
          >

          Chris,

          Farnells have part 1611790, a 24.576MHz surface mount component at
          £0.192. 10 for £0.147.
          Farnells have part 1611717, a 24.576MHz a through hole component at
          £0.23. 10 for £0.23.

          Prices are + VAT, but I think they are cheap enough.

          If these would do I would be happy to do a group buy and pass them on
          at cost, or send them to you Chris. I get free courier delivery to me
          and no minimum order.

          Regards,

          John S
        • Chris
          ... IC, giving ... Thanks for the idea, but when I have used Opto s before in nixie clock designs I have found them to be both more expensive than the
          Message 4 of 12 , Jan 11, 2009
          • 0 Attachment
            --- In smartsockets@yahoogroups.com, guus.assmann@... wrote:
            >
            > Hello Chris,
            >
            > For method c you can also use 2 times the 75468 (7 transistors per chip,
            > also in SMD. And additional transistors for the rest. (or 3 times
            IC, giving
            > 21 outputs).
            >
            > But there's also a method d.
            > Use opto-couplers as the level shifters. There's 4 in one housing types.
            > And the transistors are capable of 20mA at 70V.
            > For routing this is also simple.
            > And multiplexing to get 2 tubes with one controller is also an option.
            >
            > Fading effects are very nice. If available, they should look perfect.
            >
            > BR/
            > Guus
            >

            Thanks for the idea, but when I have used Opto's before in nixie clock
            designs I have found them to be both more expensive than the
            transistor equivalent circuit, and slow to turn on or off, but this
            second point might be something to do with the fact they were 200V opto's.

            I wonder which opto's you have in mind, and how much they cost ?

            Chris
          • Chris
            ... They re certainly good crystal prices. What I don t like about this MAXIM chip is it s price, I can t seem to find it for less than $3.50 each in large
            Message 5 of 12 , Jan 11, 2009
            • 0 Attachment
              --- In smartsockets@yahoogroups.com, Quixotic Nixotic <nixotic1@...>
              wrote:
              >
              > On 11 Jan 2009, at 10:34, Chris wrote:
              > > To improve on that, I clocked the pic from an external 24MHz crystal,
              > > to get the SPI bus up to 6MHz, and now the fading is very smooth. The
              > > problem here is that the cost of the crystal could be more than you
              > > might pay just for one tube !
              > >
              >
              > Chris,
              >
              > Farnells have part 1611790, a 24.576MHz surface mount component at
              > £0.192. 10 for £0.147.
              > Farnells have part 1611717, a 24.576MHz a through hole component at
              > £0.23. 10 for £0.23.
              >
              > Prices are + VAT, but I think they are cheap enough.
              >
              > If these would do I would be happy to do a group buy and pass them on
              > at cost, or send them to you Chris. I get free courier delivery to me
              > and no minimum order.
              >
              > Regards,
              >
              > John S
              >

              They're certainly good crystal prices. What I don't like about this
              MAXIM chip is it's price, I can't seem to find it for less than $3.50
              each in large numbers. This design also needs one PIC16F687 per tube
              too. So the final cost is based on 1of IV-4 , 1of PIC16F687 , 1of
              crystal, and 1 of PCB with interconnections. This works out at about
              £4 per tube.

              I like the idea of one PIC controlling two tubes because it cuts down
              on cost and intersocket wiring, but it's success is subject to finding
              the right driver. Guus has pointed out the 75468 , and I've looked at
              this chip already but wanted to get a feel for what others thought
              about wasting power using this driver.

              Going for the wasted power idea and three 75468 IC's at 41p each, the
              only other thing required would be a single 28-pin PIC (with internal
              oscillator)to drive two tubes, including the pcb this could cost as
              little as £1.80 per tube. The pcbs could be for two tubes but only one
              needs to be fitted if desired.

              It's just the idea of wasting power which concerns me, but it is only
              a tiny amount of power at the end of the day. I'm using a 24V supply
              wound up to 25.5V to provide the anode voltage and the total current
              for 1 tube with 8 segments turned on is 10mA, 255mW of power per tube.
              4 tubes only require 1W at 24V !

              As far as cost goes this last idea is the way to go but it isn't
              particularly elegant, but it is cheap and could produce a high quality
              multiplexed display.

              Is anybody else at odds with wasting power ?
              Chris
            • John Rehwinkel
              ... Surplus is your friend. Electronics Goldmine has 24MHz crystals for a dollar apiece. As they re surplus, they may not be the most accurate crystals out
              Message 6 of 12 , Jan 11, 2009
              • 0 Attachment
                > To improve on that, I clocked the pic from an external 24MHz crystal,
                > to get the SPI bus up to 6MHz, and now the fading is very smooth. The
                > problem here is that the cost of the crystal could be more than you
                > might pay just for one tube !

                Surplus is your friend. Electronics Goldmine has 24MHz crystals for a
                dollar apiece. As they're surplus, they may not be the most accurate
                crystals out there, but accuracy doesn't matter here. Goldmine
                usually deals in quantity, so I suspect they have a large number
                available, and may even give a discount on a quantity purchase.

                http://www.goldmine-elec-products.com/prodinfo.asp?number=G7147

                – John
              • guus.assmann@wolmail.nl
                Wasting power is not good, as a general rule. But in a case like this, it s such a small amount. And one could also say that powering up any smartsocket is a
                Message 7 of 12 , Jan 11, 2009
                • 0 Attachment
                  Wasting power is not good, as a general rule.
                  But in a case like this, it's such a small amount.
                  And one could also say that powering up any smartsocket is a waste of power....

                  BR/
                  Guus

                  >-- Oorspronkelijk bericht --
                  >To: smartsockets@yahoogroups.com
                  >From: "Chris" <fixitsan@...>
                  >Date: Sun, 11 Jan 2009 16:11:24 -0000
                  >Subject: [smartsockets] Re: IV-17 / IV-4 Smartsockets
                  >Reply-To: smartsockets@yahoogroups.com
                  >
                  >
                  >--- In smartsockets@yahoogroups.com, Quixotic Nixotic <nixotic1@...>
                  >wrote:
                  >>
                  >> On 11 Jan 2009, at 10:34, Chris wrote:
                  >Is anybody else at odds with wasting power ?
                  >Chris
                  >
                • guus.assmann@wolmail.nl
                  Hello Chris, For this, I ll have a look tomorrow. But I guess even a 4N27 or so would do just fine. BR/ Guus
                  Message 8 of 12 , Jan 11, 2009
                  • 0 Attachment
                    Hello Chris,

                    For this, I'll have a look tomorrow.
                    But I guess even a 4N27 or so would do just fine.

                    BR/
                    Guus
                    >-- Oorspronkelijk bericht --
                    >To: smartsockets@yahoogroups.com
                    >From: "Chris" <fixitsan@...>
                    >Date: Sun, 11 Jan 2009 15:50:59 -0000
                    >Subject: Re: Betr: [smartsockets] IV-17 / IV-4 Smartsockets
                    >Reply-To: smartsockets@yahoogroups.com
                    >
                    >
                    >I wonder which opto's you have in mind, and how much they cost ?
                    >
                    >Chris
                    >
                  • Quixotic Nixotic
                    Does anyone know why the IV-17s are better than the IV-4s? What is the difference, electronically-speaking? John S
                    Message 9 of 12 , Jan 11, 2009
                    • 0 Attachment
                      Does anyone know why the IV-17s are better than the IV-4s? What is
                      the difference, electronically-speaking?

                      John S
                    • Chris
                      ... Sorry John, but the accuracy is important because the crystal is also the timebase for the USART serial port, although I am not sure to what extent a few
                      Message 10 of 12 , Jan 11, 2009
                      • 0 Attachment
                        --- In smartsockets@yahoogroups.com, John Rehwinkel <jrehwin@...> wrote:
                        >
                        > > To improve on that, I clocked the pic from an external 24MHz crystal,
                        > > to get the SPI bus up to 6MHz, and now the fading is very smooth. The
                        > > problem here is that the cost of the crystal could be more than you
                        > > might pay just for one tube !
                        >
                        > Surplus is your friend. Electronics Goldmine has 24MHz crystals for a
                        > dollar apiece. As they're surplus, they may not be the most accurate
                        > crystals out there, but accuracy doesn't matter here. Goldmine
                        > usually deals in quantity, so I suspect they have a large number
                        > available, and may even give a discount on a quantity purchase.
                        >
                        > http://www.goldmine-elec-products.com/prodinfo.asp?number=G7147
                        >
                        > – John
                        >

                        Sorry John, but the accuracy is important because the crystal is also
                        the timebase for the USART serial port, although I am not sure to what
                        extent a few ppm of crystal error at 24MHz changes the port speed.
                        I had some issues with internal oscillators running at 4MHz being
                        slightly inaccurate.
                      • Chris
                        ... of power.... ... Yeah, but hopefully they won t ever say that ! I ve just done some real measurements and had a pleasant surprise. At 24V the long diagonal
                        Message 11 of 12 , Jan 11, 2009
                        • 0 Attachment
                          --- In smartsockets@yahoogroups.com, guus.assmann@... wrote:
                          >
                          > Wasting power is not good, as a general rule.
                          > But in a case like this, it's such a small amount.
                          > And one could also say that powering up any smartsocket is a waste
                          of power....
                          >

                          Yeah, but hopefully they won't ever say that !

                          I've just done some real measurements and had a pleasant surprise.
                          At 24V the long diagonal segment (chosen because it was the longest)
                          runs at about 140uA (microamps). As voltage rises so too will current,
                          but it's probably enough for now to say that each segment draws 200uA,
                          to be on the safe side.

                          If , on average, only half of the segments are used, the average
                          maximum current lost in the driver will be no more than about 2mA.

                          Compared to a 'proper' VFD high side driver like the MAX6921, which
                          draws about 1mA in idle mode, and has serial bus lines and blanking
                          control lines which could easily add 10mA, using resistors and drivers
                          seems to be less wasteful than a proper VFD driver is !(under certain
                          circumstances ;o) )

                          Now, looking again at the choice of driver IC's can anyone see
                          anything wrong with using three ULN2003AD chips at 15 pennies each ?
                          (less in larger quantities). They are rated to 50V with a relatively
                          huge current capacity. I have some on the shelf so I think I'll give
                          them a try in the next couple days.

                          The other thing missing from the equation are the resistor networks
                          which would be required to add current limiting to each segment which
                          is to be shorted to ground, just pennies in terms of cost, but need to
                          find room for them on the pcb too !

                          Chris
                        • Chris
                          ... Apparently they live longer, and also draw slightly less current too.
                          Message 12 of 12 , Jan 11, 2009
                          • 0 Attachment
                            --- In smartsockets@yahoogroups.com, Quixotic Nixotic <nixotic1@...>
                            wrote:
                            >
                            > Does anyone know why the IV-17s are better than the IV-4s? What is
                            > the difference, electronically-speaking?
                            >
                            > John S
                            >


                            Apparently they live longer, and also draw slightly less current too.
                          Your message has been successfully submitted and would be delivered to recipients shortly.