Stef wrote:> Mike Harrison <mike@whitewing.co.uk> wrote: > >>On Sat, 12 Jun 2004 13:29:51 GMT, Mostafa Kassem <No_one@Nowhere.com> wrote: >>>Stef wrote: >>> >>>>Hi all, >>>> >>>>I am searching for a driver for 6-8 digits of 5x7 LED matrix displays. >>>>All I can find seems to be the 4-digit MAX6952/3 chips. Two of these >>>>would ofcourse work, but that requires a lot of space and they are not >>>>cheap either. I do not really need all the font functionality etc. As >>>>long as I can control each LED seperately, I'm OK. Another option would >>>>be a chip with only the drive transistors and current sources and do >>>>the multiplexing in SW. >>>> >> >>Probably the cheapest way would be to use a string of 74HC595s as column drivers and transistors >>(maybe driven from a HC138 if I/O is tight) as row drivers, and use a timer interrupt task to keep >>them refreshed. > > > Yes, this is an option, but the drive capability of the 74HC595 may be a bit > on the low side and I may need to add a current source somewhere.There are two paths on this one : Buy a device that handles current sink, and has define on current values, such as http://www.maxim-ic.com/quick_view2.cfm/qv_pk/4206 Web indicates $1.95 for 17 nodes, very small package (etc) Or use vanilla logic, and resistors, eg there are 595's with more muscle, such as : http://focus.ti.com/docs/prod/folders/print/tpic6b595.html web indicates 49c for 8 nodes, higher drive than above, so bigger package. I think ST also have these power driver '595's. If you work back from the current level and control needed, decide if single-source is an issue, and that will narrow your choices. -jg
Searching for 8-digit 5*7 LED matrix display driver
Started by ●June 11, 2004
Reply by ●June 14, 20042004-06-14
Reply by ●June 15, 20042004-06-15
In comp.arch.embedded, Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote:>On Mon, 14 Jun 2004 22:35:07 +1000, the renowned dmm!@work.now wrote: > >>On Fri, 11 Jun 2004 15:42:08 +0200, stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote: >> >>>Hi all, >>> >>>I am searching for a driver for 6-8 digits of 5x7 LED matrix displays. >>>All I can find seems to be the 4-digit MAX6952/3 chips. Two of these >>>would ofcourse work, but that requires a lot of space and they are not >>>cheap either. I do not really need all the font functionality etc. As >>>long as I can control each LED seperately, I'm OK. Another option would >>>be a chip with only the drive transistors and current sources and do >>>the multiplexing in SW. >> >>Then again there's TI's TLC5920, a 16 x 8 LED driver. $2.90 ea 1K+. Samples are available. >> >>If you wish to drive the rows and columns with seperate ICs, then you can drive the columns >>with a TI TPIC6595 8-bit serial shift register which has power DMOS switches. > >The rows can be driven with 3 dual p-channel MOSFETs. > >Parts count: 7 xx595s SMT > 7 8-resistor arrrays (or 14 quads) SMT > 3 dual p-channel MOSFETS SO-8 > --- > 17/24 parts >Yes, this can work. I suppose you would drive this in a 1/5 scan ratio and limit the peak current to the '595's drive capability? Downside is that there is no current regulation and especially with blue LED's (Vfwd ~= 3.5V) this can cause significant differences between LED's. To solve this I'd like to use current drivers. So rearranging the martrix a little and using a scan ratio of 1/14 (as does the MAX6952) and a high(er) peak current I come to this: 1 MBI5027 16 channel current sink 2 8 channel high-side driver, or 7 dual P-FET, or 14 singles 2 optionally '595's if there is not enough I/O to drive the P's --- 3-17 parts This gives me exactly enough drivers to drive 6 digits and an extra column of 14 discrete LED's. And the already scanning P-FET's allow for easy scanning of a keypad. But this TLC5920 also looks nice, it may decrease the part count to 2. -- Stef (remove caps, dashes and .invalid from e-mail address to reply) "Money is the root of all money." -- the moving finger
Reply by ●June 15, 20042004-06-15
In comp.arch.embedded, Jim Granville <no.spam@designtools.co.nz> wrote:>Stef wrote: >> Mike Harrison <mike@whitewing.co.uk> wrote: >> >>>Probably the cheapest way would be to use a string of 74HC595s as column drivers and transistors >>>(maybe driven from a HC138 if I/O is tight) as row drivers, and use a timer interrupt task to keep >>>them refreshed. >> >> >> Yes, this is an option, but the drive capability of the 74HC595 may be a bit >> on the low side and I may need to add a current source somewhere. > >There are two paths on this one : > Buy a device that handles current sink, and has define on current >values, such as >http://www.maxim-ic.com/quick_view2.cfm/qv_pk/4206 >Web indicates $1.95 for 17 nodes, very small package (etc) > > Or use vanilla logic, and resistors, >eg there are 595's with more muscle, such as : >http://focus.ti.com/docs/prod/folders/print/tpic6b595.html >web indicates 49c for 8 nodes, higher drive than above, so bigger >package. I think ST also have these power driver '595's. >As you may have seen in other posts I am leaning towards special purpose current sinks. These are simular to '595's but have current regulated outputs. -- Stef (remove caps, dashes and .invalid from e-mail address to reply) The man who runs may fight again. -- Menander
Reply by ●June 15, 20042004-06-15
On Tue, 15 Jun 2004 17:01:05 +0200, the renowned stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote:>In comp.arch.embedded, >Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote: >>On Mon, 14 Jun 2004 22:35:07 +1000, the renowned dmm!@work.now wrote: >> >>>On Fri, 11 Jun 2004 15:42:08 +0200, stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote: >>> >>>>Hi all, >>>> >>>>I am searching for a driver for 6-8 digits of 5x7 LED matrix displays. >>>>All I can find seems to be the 4-digit MAX6952/3 chips. Two of these >>>>would ofcourse work, but that requires a lot of space and they are not >>>>cheap either. I do not really need all the font functionality etc. As >>>>long as I can control each LED seperately, I'm OK. Another option would >>>>be a chip with only the drive transistors and current sources and do >>>>the multiplexing in SW. >>> >>>Then again there's TI's TLC5920, a 16 x 8 LED driver. $2.90 ea 1K+. Samples are available. >>> >>>If you wish to drive the rows and columns with seperate ICs, then you can drive the columns >>>with a TI TPIC6595 8-bit serial shift register which has power DMOS switches. >> >>The rows can be driven with 3 dual p-channel MOSFETs. >> >>Parts count: 7 xx595s SMT >> 7 8-resistor arrrays (or 14 quads) SMT >> 3 dual p-channel MOSFETS SO-8 >> --- >> 17/24 parts >> >Yes, this can work. I suppose you would drive this in a 1/5 scan ratio and >limit the peak current to the '595's drive capability?If you use the DMOS open-drain XX595's then you can drive up to a couple hundred mA per output depending on version (40mA average per LED, probably too much). You would probably have to upgrade the p-channel transistors if you go that high (the total power dissipation of display + drivers + resistors could be as high as 20~30W.>Downside is that there is no current regulation and especially with blue >LED's (Vfwd ~= 3.5V) this can cause significant differences between LED's.Current can be significantly different with no visual effect. I think you'll find it's perfectly acceptable.>To solve this I'd like to use current drivers. So rearranging the martrixIf your average current requirements are rather modest and don't mind a fair bit of dissipation in the semiconductors and don't mind using single-sourced parts then it indeed might be the right solution for you.>a little and using a scan ratio of 1/14 (as does the MAX6952) and a >high(er) peak current I come to this: > > 1 MBI5027 16 channel current sink > 2 8 channel high-side driver, or 7 dual P-FET, or 14 singles > 2 optionally '595's if there is not enough I/O to drive the P's > --- > 3-17 parts > >This gives me exactly enough drivers to drive 6 digits and an extra >column of 14 discrete LED's. And the already scanning P-FET's allow >for easy scanning of a keypad. > >But this TLC5920 also looks nice, it may decrease the part count to 2.Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Reply by ●June 15, 20042004-06-15
Stef wrote:> Hi all, > > I am searching for a driver for 6-8 digits of 5x7 LED matrix displays. > All I can find seems to be the 4-digit MAX6952/3 chips. Two of these > would ofcourse work, but that requires a lot of space and they are not > cheap either. I do not really need all the font functionality etc. As > long as I can control each LED seperately, I'm OK. Another option > would > be a chip with only the drive transistors and current sources and do > the multiplexing in SW.I use the SN74LVTH573-PW from Ti for this, as my micro is an MSP430 (Vdd <= 3V6). It has 5V tolerant outputs and 64mA/128mA high/low maximum current capability. I multiplex the parts using the /OE pin, for brightness control, and current limiting, and add current limiting resistors where desirable. The 573 makes connection simple as it has the input bus on the low pin number side, and the output bus on the higher numbered pins, thus allowing a straight through data bus. It is expensive in real estate terms, requiring 9 -PW (SSOP) packages per 4 LED blocks, but these are small and extremely cheap Cheers Al
Reply by ●June 16, 20042004-06-16
In comp.arch.embedded, Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote:>On Tue, 15 Jun 2004 17:01:05 +0200, the renowned >stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote: >>> >>Yes, this can work. I suppose you would drive this in a 1/5 scan ratio and >>limit the peak current to the '595's drive capability? > >If you use the DMOS open-drain XX595's then you can drive up to a >couple hundred mA per output depending on version (40mA average per >LED, probably too much). You would probably have to upgrade the >p-channel transistors if you go that high (the total power dissipation >of display + drivers + resistors could be as high as 20~30W. >Ah, never thought of high current 595's, thanks.>>Downside is that there is no current regulation and especially with blue >>LED's (Vfwd ~= 3.5V) this can cause significant differences between LED's. > >Current can be significantly different with no visual effect. I think >you'll find it's perfectly acceptable. >Do you have any experience with this? I've seen some websites saying that current is important, but those where of current source manufacturers. ;-)>>To solve this I'd like to use current drivers. So rearranging the martrix >I've now got a qoute of under EUR 1.00 (1000+) for a 16 channel current sink. This would replace 2 high current 595's and 16 resistors. So costwise current regulation is not bad. Not even single source looks like a problem, the Allegro and Macroblock parts are pin compatible. I'll have to compare them in more detail to see if they are fully drop-in replacements. For now I think I have enough info to select the best solution, pity the ideal single-chip solution did not pop up. Thanks to all who replied. -- Stef (remove caps, dashes and .invalid from e-mail address to reply) I hope you millionaires are having fun! I just invested half your life savings in yeast!!
Reply by ●June 17, 20042004-06-17
On Wed, 16 Jun 2004 12:22:43 +0200, stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote:>In comp.arch.embedded, >Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote: >>On Tue, 15 Jun 2004 17:01:05 +0200, the renowned >>stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote: >>>> >>>Yes, this can work. I suppose you would drive this in a 1/5 scan ratio and >>>limit the peak current to the '595's drive capability? >> >>If you use the DMOS open-drain XX595's then you can drive up to a >>couple hundred mA per output depending on version (40mA average per >>LED, probably too much). You would probably have to upgrade the >>p-channel transistors if you go that high (the total power dissipation >>of display + drivers + resistors could be as high as 20~30W. >> >Ah, never thought of high current 595's, thanks. > >>>Downside is that there is no current regulation and especially with blue >>>LED's (Vfwd ~= 3.5V) this can cause significant differences between LED's. >> >>Current can be significantly different with no visual effect. I think >>you'll find it's perfectly acceptable. >> >Do you have any experience with this? I've seen some websites saying >that current is important, but those where of current source >manufacturers. ;-) > >>>To solve this I'd like to use current drivers. So rearranging the martrix >> > >I've now got a qoute of under EUR 1.00 (1000+) for a 16 channel current >sink. This would replace 2 high current 595's and 16 resistors. So costwise >current regulation is not bad. Not even single source looks like a problem, >the Allegro and Macroblock parts are pin compatible. I'll have to compare >them in more detail to see if they are fully drop-in replacements. > >For now I think I have enough info to select the best solution, pity the >ideal single-chip solution did not pop up. > >Thanks to all who replied.I requested 4 samples of Texas Instruments' TLC5920DL two nights ago, and I received them this afternoon. I will be playing with them as soon as the boards are returned from the pcb manufacturer. (They have to be designed yet !!) I have nothing but praise for TI's response time.
Reply by ●June 17, 20042004-06-17
On Wed, 16 Jun 2004 12:22:43 +0200, the renowned stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote:>In comp.arch.embedded, >Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote: >>On Tue, 15 Jun 2004 17:01:05 +0200, the renowned >>stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote: >>>> >>>Yes, this can work. I suppose you would drive this in a 1/5 scan ratio and >>>limit the peak current to the '595's drive capability? >> >>If you use the DMOS open-drain XX595's then you can drive up to a >>couple hundred mA per output depending on version (40mA average per >>LED, probably too much). You would probably have to upgrade the >>p-channel transistors if you go that high (the total power dissipation >>of display + drivers + resistors could be as high as 20~30W. >> >Ah, never thought of high current 595's, thanks.TI has at least two versions with different drive current capability (and pinout, unfortunately). Single-sourced, AFAIK, but they have been around for about a decade.>>>Downside is that there is no current regulation and especially with blue >>>LED's (Vfwd ~= 3.5V) this can cause significant differences between LED's. >> >>Current can be significantly different with no visual effect. I think >>you'll find it's perfectly acceptable. >> >Do you have any experience with this? I've seen some websites saying >that current is important, but those where of current source >manufacturers. ;-)Rather similar situations. I don't think it's a problem.>>>To solve this I'd like to use current drivers. So rearranging the martrix >> > >I've now got a qoute of under EUR 1.00 (1000+) for a 16 channel current >sink. This would replace 2 high current 595's and 16 resistors. So costwise >current regulation is not bad. Not even single source looks like a problem, >the Allegro and Macroblock parts are pin compatible. I'll have to compare >them in more detail to see if they are fully drop-in replacements.That sounds reasonable. What part did you get the $1.20-ish quote on? The TI parts are more like $4 USD in 100's. Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Reply by ●June 17, 20042004-06-17
In comp.arch.embedded, Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote:>On Wed, 16 Jun 2004 12:22:43 +0200, the renowned >stef33d@yahooI-N-V-A-L-I-D.com.invalid (Stef) wrote: >> >>I've now got a qoute of under EUR 1.00 (1000+) for a 16 channel current >>sink. This would replace 2 high current 595's and 16 resistors. So costwise >>current regulation is not bad. Not even single source looks like a problem, >>the Allegro and Macroblock parts are pin compatible. I'll have to compare >>them in more detail to see if they are fully drop-in replacements. > >That sounds reasonable. What part did you get the $1.20-ish quote on? >The TI parts are more like $4 USD in 100's. >This was a qoute for the MBI5027. This was for another project, which I am not involved in, so I don't know the exact numbers. but it was at least 1000+, maybe 5k. -- Stef (remove caps, dashes and .invalid from e-mail address to reply by mail) Kime's Law for the Reward of Meekness: Turning the other cheek merely ensures two bruised cheeks.
