Ice Tube Clock IV-18 To-Spec Hack

For RTC breakouts, etc., use the Other Products from Adafruit forum

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

I thought it might be fun to run the Ice Tube Clock's IV-18 tube to specifications. With all the additional wires, this mod is perfect for all you hackers who would like your clocks to look like a spaghetti creature wormed its way beneath the circuit boards and died.

The IV-18 specifications recommends 50 volts for anode/grid pulses, with an absolute maximum of 70 volts. Setting OCR0A to 128 in the software is sufficient to provide 55-60 volts, and the clock's Zener diode, D5, prevents voltage from ever exceeding 60 volts.

The Ice Tube Clock normally controls brightness by changing the anode/grid voltage, so with a constant voltage, another method is required for controlling display brightness. The most straightforward way is applying a PWM signal to the MAX6921 BLANK pin. By rapidly enabling and disabling each digit, the display will appear dimmer and brightness can be controlled by changing the duty cycle of the PWM signal.

The microcontroller provides three timers capable of generating PWM signals, but all are already used for various functions. Even so, timer 0 supplies a PWM signal to the boost circuit and is active whenever the display is active. Timer 0 can also be configured to output a second PWM signal on PD5/OC0B. Inconveniently, PD5 is already used for the MENU button, so modifications are required:
  • cut trace between PD5 and the MENU button
  • cut trace between PC3 and the MAX6921 BLANK pin
  • wire PB4 to the MENU button (green)
  • wire PD5 to the MAX6921 BLANK pin (blue)
The firmware is then configured to query PB4 for MENU button input, and timer 0 is configured to output a suitable PWM signal on PD5/OC0B:

Code: Select all

// COM0A1:0 = 10: clear OC0A on compare match; set at BOTTOM
// COM0B1:0 = 11: clear OC0B at bottom; set on compare match
// WGM02:0 = 011: clear timer on compare match; TOP = 0xFF
TCCR0A = _BV(COM0A1) | _BV(COM0B0) | _BV(COM0B1) | _BV(WGM00) | _BV(WGM01);
The IV-18 specifications recommends 5 volts for the cathode (heater), with an acceptable range of 4.3 to 5.5 volts. The Ice Tube Clock normally uses a MOSFET to drive the filament with an occasionally inadequate 3-4 volts of direct current. A frequent symptom of this underperformance is a dim digit on the display, but the resulting problems can be more dramatic.

VFDs are typically driven by alternating current, so to generate an alternating current signal from the microcontroller, PC2 and PC3 are configured as outputs, with PC2 initially pushed high and PC3 initially pulled low:

Code: Select all

DDRC  |=  _BV(PC2);  DDRC  |=  _BV(PC3);
PORTC |=  _BV(PC2);  PORTC &= ~_BV(PC3);
The timer 0 overflow interrupt is called every 32 microseconds, and in this interrupt, the PC2 and PC3 output values are reversed. Thus the voltage potential between PC2 and PC3 alternates between 5 and -5 volts with a frequency of around 17 kHz:

Code: Select all

PORTC ^= _BV(PC2) | _BV(PC3);
Inconveniently, the microcontroller pins can only output a maximum 20 mA, but according to the datasheet, the IV-18 could consume up to 95 mA. So two push-pulls constructed with power transistors (PN2222 and PN2907A; pictured below) amplify the alternating current signal. A TSC428 MOSFET might have been a better choice, but I happened to have suitable power transistors on hand.

ImageImage

The push-pull circuit is incorporated into the clock as follows:
  • glue the push-pull circuit board to the right plexiglass sidepiece
  • cut traces from ground to female header pins 11 and 12 (picture below)
  • desolder and remove R3
  • wire the first push-pull input to microcontroller pin PC2 (yellow)
  • wire the second push-pull input to microcontroller pin PC3 (yellow)
  • wire the push-pull power to the clock power (red)
  • wire the push-pull ground to the clock ground (black)
  • wire the first push-pull output to female header pin 2 (white)
  • wire the second push-pull output to female header pins 11 and 12 (white)
ImageImage

The full-board photo below is linked to a larger image, but please ignore the rightmost white wire. It's part of an unrelated mod to extend battery life.

Image

Overall, I'm satisfied with the end result, pictured below. At very low brightness, individual segments seem to fluoresce more evenly. Increased voltage to the cathode eliminates the dim digit problem exhibited by some displays. And there is no brightness gradient across the display, which was sometimes noticeable in the original design at very low brightness (albeit barely). All-in-all, the improvement is minor, but there is something aesthetically pleasing about driving the IV-18 "to spec."

Image

The software portion of this hack is implemented in my unofficial xmas-icetube firmware and may be downloaded from GitHub.

User avatar
wildtang3nt
 
Posts: 35
Joined: Thu Aug 09, 2012 2:31 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by wildtang3nt »

This looks like an awesome hack, maybe a little ambitious for me, but I definitely agree that there's something to be said for doing it right :)

One of these days when I learn how to use Eagle, I am going to take all the hardware improvements you and others have devised and do a re-spin of the PCB. I don't mind the dead multi-coloured spider look, but I prefer the clean look I guess, and it would be a challenge to me. Adafruit will probably get around to it before I do though lol

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

One of these days when I learn how to use Eagle, I am going to take all the hardware improvements you and others have devised and do a re-spin of the PCB.
That would be awesome! I thought about doing the same, but did not have the time to lean Eagle. If you do decide to improve upon the Adafruit design, it might be a good idea to start a new thread to solicit ideas from existing users.
Adafruit will probably get around to it before I do though lol
Somehow I don't think that will happen. The idea of a redesign has been brandied about for many years on this forum, so there is definitely interest. But I haven't seen anything come of it. It might have something to do with the finite supply of IV-18 tubes...

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

jarchie wrote: A TSC428 MOSFET might have been a better choice...
I tried the equivalent of a TSC438 MOSFET--a MAX627 driver. Unfortunately, the voltage drop was too great--insufficient for the the IV-18 tube.
Russell 27 wrote:Often this type of display is driven with alternating current. But true sine wave alternating current is different than a push pull circuit, producing a modified, square sine wave. Can't say the display will notice this difference.
I copied the above quote from another thread, but I wanted to state here that Russell 27 is 100% correct. I am indeed generating square wave AC when the original IV-18 was, no doubt, designed for sine wave AC.

Remarkably, Russell 27 correctly inferred the reason I erroneously equivocated square wave AC with sine wave AC: I believe the tube will be ambivalent to the square-versus-sine wave difference, an intuition I draw from the Noritake guide.

The bottom line is that anyone using this "to-spec" hack should be aware that my use of a push pull circuit is indeed generating square wave AC which is not what the designers of the IV-18 intended for the filament.

Thank you, Russell 27, for pointing out my oversight!

User avatar
phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by phild13 »

This is stretching what I remember so read the following opinions accordingly.

Though probably most tubes used to have AC on their filaments (the filament winding tap from the power transformer) and were designed for that, I wouldn't see any real reason not to drive the filament with a square wave.

A pure AC sine I think would probably give the best evenness in brightness across multiple segments of the tube, and generally would not cause any cooling of the filament.

Half wave AC might cause some filament cooling due to the time period the current is off (about 8.3 mS for 60 hz I think) and may increase the thermal stress on the filament a bit as a result.

Pure DC would probably provide the best filament life with no thermal cycling once turned on. The stress would be the initial application of current to the filament which I think would be greater than with AC. Driving the filament with straight DC will cause an uneven brightness across the tubes display. How noticeable this brightness change is probably depends on the individual tubes construction tolerance, the person observing, and ambient lighting.

A 50% duty cycle square wave I think would fall somewhere in between the two with a higher frequency (shorter pulse time) better. We have to watch the frequency though or we may cause too much interference. AC is probably the easiest on the filament overall, but I would think that if the square wave frequency were close to what the filament design expects or higher then thermal cycling between positive and negative pulses would be within design specs of the filament even though it is not AC. I would expect similar tube brightness evenness between digits as would be found in an AC sine.

In today's digital world mostly void of tube equipment with associated big tapped power transformers and/or dedicated filament transformers, generating a square wave by some method to drive a tube filament is probably the most common and easiest option for driving VFD's and other tubes.

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

Thank you for the comments!

For reference, the approach described in the original post generates square wave AC with a frequency in the tens of kilohertz range, but the wavelength is inconsistent: interrupt blocking prevents the AC-generation code from running at perfectly regular intervals.
PhilD13 wrote:This is stretching what I remember so read the following opinions accordingly.
My knowledge is a bit stretched here also, so the same caveat applies to me.
PhilD13 wrote:A pure AC sine I think would probably give the best evenness in brightness across multiple segments of the tube, and generally would not cause any cooling of the filament.
With sine wave AC, the voltage potential between filament ends gradually alternates between positive and negative. When the potential is near zero less current flows. But when the potential is at a peak or trough, more current flows. Naively, I would imagine that the filament would warm and cool proportionally to the magnitude of the sine wave.
PhilD13 wrote:Half wave AC might cause some filament cooling due to the time period the current is off (about 8.3 mS for 60 hz I think) and may increase the thermal stress on the filament a bit as a result.
Sounds right to me.
PhilD13 wrote:Pure DC would probably provide the best filament life with no thermal cycling once turned on. The stress would be the initial application of current to the filament which I think would be greater than with AC.
I agree that thermal stress is minimized with direct current, but another failure mode, DC notching, is a significant problem when driving filaments with DC. Another forum user, neutron spin, pointed me to the most definitive report I've seen so far: WRDC-TR-90-4075.

Tungsten filaments are composed of dense and loose regions. The densely packed regions are crystalline granules with a cubic structure. The loosely packed regions lack definite structure and may also contain trace contaminates. The thermionic effects of heating the filament not only emit electrons but also generate ions. Due to the consistent electric field under DC current, these ions move in only one direction within the filament, a process called electromigration. The ions move more easily in the loosely packed regions of tungsten. Eventually electromigration moves material away from loosely packed regions, creating notches in the filament.

Notched regions of filament contain less metal, but must carry the same current. As the notches become hotter the rate of electromigration and sublimation increase, deepening the notches at an increasing rate. Eventually the notched areas become so thin and hot that they break under normally insignificant mechanical stress.

I don't think any metal has a perfectly uniform structure, so DC notching should affect other filament chemistries to a greater or lesser degree. Even so, I also don't believe that filament notching will effect the life of the Ice Tube Clock, as argued in another thread. I predict that DC-driven IV-18 tubes will fail from phosphor degradation long before the effect of filament notching causes a filament failure.
PhilD13 wrote:A 50% duty cycle square wave I think would fall somewhere in between the two with a higher frequency (shorter pulse time) better. We have to watch the frequency though or we may cause too much interference.
Naively, I would think that square wave AC would be better than standard AC because the current across the filament is constant--only the direction changes. I also think that square wave AC, like sine wave AC, will prevent DC notching.

But some electrons are lost as they are emitted by the filament, so there is slightly more current at the negative side than the positive side. I would expect a slight temperature gradient across the filament with the negative side being slightly hotter than the positive side. So square wave AC will still cause some thermal cycling, but as you wrote, using a high frequency should virtually eliminate the stress.

In general, high frequency square wave AC might be the most reliable way to drive VFD filaments, but I have yet to see any convincing research to support that belief.
PhilD13 wrote:In today's digital world mostly void of tube equipment with associated big tapped power transformers and/or dedicated filament transformers, generating a square wave by some method to drive a tube filament is probably the most common and easiest option for driving VFD's and other tubes.
When since wave AC is not readily available, my understanding is that high frequency square wave AC is the most common way to drive VFD filaments.

User avatar
phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by phild13 »

I should have said square wave AC and not just square wave as I was meaning a square wave AC and not a series of positive square wave DC pulses. Sometimes what is in the mind (or not in it) and what the fingers do are different. What I am calling duty cycle is the pulse duration on each side of zero being approximately equal to each other. I don't think a variation in duty cycle would have much effect on the actual operation of the filament. The pulses would generally be fast enough for it to not really notice.
I took a look at the link below of section 5 and it does show a variation in pulse width and amplitude between the positive side and the negative side on the 5.3 Pulse Filament Drive (High Frequency RMS) section.
http://www.noritake-elec.com/vfd_operat ... FDFilament


My understanding with the sine wave ac is that the transition from positive peak to negative peak is fast enough at 60 Hz that the filament does not have time to cool any appreciable amount.

If the power was a half wave ac sine then there is a much longer "off" period between peaks (roughly twice as long) and the filament starts to cool into the range where there will be some thermal stress.

I think with the square wave AC your are getting essentially the same as you get with a sine wave and maybe like you say even better because the rise/fall period is faster than in a sine. Though it must not be much of a concern, as I find no mention so far, I am not sure if it is better on the initial power on as the voltage and current rises very fast to the peak.

I forgot all about Dc notching of filaments and even have an Allen Bradley pdf document that discusses illumination selection for indicator lights which I used a few years ago to point out why I thought the motors on lamps in the Robots at work were failing so often. (24 volts DC powered).

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

I appreciate the clarification and comments, and what you wrote makes sense to me and is consistent with the small bit of research I did before posting this hack. Although I am good at doing research and thinking logically, I have very little experience in electronics. So I definitely feel better about my contributions knowing that experienced engineers, like yourself, are viewing my posts. :-)

Thank you!

User avatar
russell 27
 
Posts: 242
Joined: Thu Sep 12, 2013 3:59 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by russell 27 »

One other thing about AC that comes from a transformer via a generator or alternator, is the way the wave form is produced and peak voltage. A square wave is either on or off for a duration, and particular voltage, in either polarity. If you buy one of those inexpensive 12 volt dc to 120 volt AC converters this is what you get, generally called a modified sine wave. Inductive type circuits usually don't work well with this type of wave form. True sine wave is a nice even, swooping arc turning on and off, as the polarity in the generator turns through it's cycle. Where a DC measurement from your meter shows actual voltage, an AC measurement from meter is actually RMS, which stands for root means squared. To avoid an unwanted explanation, RMS is actually kind of an average. Since a great deal of the actual time, AC is either turning on or off, and only at the peak ON for a fraction of the time, the RMS is lower than peak voltage. to find peak from RMS multiply by 1.414. So the peak voltage of 5 VAC is 7.07 volts. This may not seem like much, but at 120 vac the peak is 170 volts. That's why sometimes when you switch on a filament light bulb, you hear that clink, BOOM, and the lamp burns out. When a lamp filament is at rest there is no resistance, when you switching on the lamp it immediately heats up and makes light, but in doing so the heated filament makes resistance and now regulates current. If you happen to throw the switch right at the point peak voltage is 170 volts, often an older weak filament can't handle that anymore and it burns out. Of course you may also throw the switch when the sine is at zero volts. Sorry I couldn't help it, this kind of goes along with all the other discussion.

User avatar
phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by phild13 »

I'm going to try to get this explained right as I dislike math, but is always seems to pop up.....

Actually all waveforms have a rms value (average value), a peak value, and in the case of an alternating signal, a Peak to Peak value. This can be seen in the image below.

As implied by the name, voltage rms (Vrms) is calculated by taking the square root of the mean average of the square of the voltage in a chosen interval.

In calculating non-sinusoidal signals such as a square wave, it is usually acceptable to choose the first cycle. Calculations are often represented in radians. The waveform goes from 0 radians which is 0° through π/2 radians which is 90°. The waveform repeats every 2π radians or 360°. Voltage and current exhibiting cyclic behavior is referred to as alternating. One full cycle Ac square wave is shown.

There are a number of ways in which the square wave can be referenced, peak voltage, peak-to-peak voltage, average voltage, and root-mean-square voltage (Vrms). Peak voltage and peak-to-peak voltage are apparent by looking at the above plot. Root-mean-square and average voltage are not so apparent and need to be calculated.

Without getting into a bunch of calculations which I'm likely to get wrong anyway, you can think of it this way. The negative pulse contributes energy, so both the positive and negative pulses need to get calculated. The percentage of duty cycle does not matter as duty cycle refers to length of on to off and the amplitude of the signal is the same for the waveform. The total rms value of the waveform would then be calculated by applying the square root of the sum of squares of the two individual rms values. An oscilloscope provides a great look at peak to peak of a signal and it is easiest to determine what the value is from a scope.
Example of Square wave values
Example of Square wave values
square_wave_rms_avg.gif (3.35 KiB) Viewed 4918 times
For the light bulb, there is a very low resistance to a cold filament. A 100 watt 120 volt incandescent bulb has a hot resistance of about 144 ohms (0.83A) when in operation, and a cold resistance that is much lower at about 9.5 ohms (12.63A). When the rated RMS voltage is applied to the cold filament a very large current will flow for about 0.10 second before stabilizing and reaching 90% brightness in about 0.13 of a second. this places a great stress on the filament.

As all filaments evaporate when heated during operation (there are various gases and coatings used to slow this process) and will eventually fail. Filament manufacture will also affect the life as small variations in thickness causes hotspots that evaporate faster. The filament will eventually have very thin spots that are much weaker (DC notching makes this more pronounced and quicker) than the bulk of the filament. A bulb "blows" when the weak spots eventually become too thin to withstand such a large inrush of current that gets applied to a cold filament causing a localized overheating of the filament wire in the weakest spot, resulting in melting at the weak point and seperation.

Bulbs used with a light dimmer will generally last a lot longe as the dimmer limits the inrush current to the bulb thus softening the stress the filament needs to absorb and making it less likely weak spots in the filament wire will overheat and melt.

User avatar
russell 27
 
Posts: 242
Joined: Thu Sep 12, 2013 3:59 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by russell 27 »

What he said.

Actually what I was referring to was pure DC, as opposed to AC when making meter measurements. DC has no peak and AC does. Apparently I rambled on too much. I consider knowing this about voltage important. Peak voltage hitting a low resistance filament is hard on it, If one could always turn on the light switch at the zero cross point, lamps would last longer, I've found this to be true. The filament in a bulb acts different than many resistive circuits. Not so much of a problem with a 5 volt potential.

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

Thank you both for the commentary!


On another note, this to-spec hack is, unfortunately, much less to spec than I thought...

Last weekend, I was checking the electrical characteristics of my xmas-icetube board redesign, which incorporates this to-spec hack. When viewing the AC signal on an oscilloscope, I discovered that the AC voltage across the filament is just shy of 3 volts! I've previously claimed that this hack drives the filament with around 4.6 volts; that estimate was way off.

As measured in another post, the original Adafruit design uses a MOSFET which drops voltage to 3.3v, allowing 53 mA of current across R3 and the IV-18 filament. The total resistance of R3 and the filament is 3.3/0.052 or 63 ohms. Since R3 is 22 ohms, the filament resistance is 63-22 or 41 ohms. Because R3 and the IV-18 filament form a voltage divider, the voltage across the filament is 3.3*41/(22+41) or 2.1 volts. So the to-spec hack still seems to be an improvement over the original design, but should probably be renamed the "still-far-from-spec" hack.

With the Adafruit kit, replacing R3 with a jumper provides an easy way to increase the DC voltage across the IV-18 filament to 2.89 volts, solving the dim digit issue in most cases. That's comparable to the 2.85 volts of AC delivered by the current to-spec hack. (Voltage measured with a true RMS multimeter.)

In another thread, PhilD13 mentioned the ZVP2110A and ZVN2110A MOSFETS, and after looking at the data sheets for those FETs, I think that they will result in a lower voltage drop. I've ordered those FETs from DigiKey as well as some LM317 voltage regulators. My hope is that with those parts, I can figure out how to drive the filament at 5 volts, as required by the tube specifications.

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

Simply replacing the BJTs in the original to-spec hack with ZVP2110A and ZVN2110A MOSFETs increases filament voltage to 3.64 volts--definitely an improvement, but still shy of spec.

I'd like to increase that to ~5v, but wish to do so with a minimum of parts so I can fit the circuit on a redesigned Ice Tube Clock board. So I wired an LM317 regulator to the output of D1 (~9v) and used a simple resistor divider to configure the output to ~6v for the push-pulls:
Quick scribble of the my bread-boarded filament power supply.
Quick scribble of the my bread-boarded filament power supply.
filament-driver.jpg (20.63 KiB) Viewed 4808 times
According to the LM317 datasheet, a capacitor on the output "is not needed for stability; however, it does improve transient response." So I have no capacitor on the output, and without a cap, there's also no need for a protection diode. That simple circuit provides ~5v of somewhat noisy high frequency AC.

While the push-pulls were driving an IV-18 filament, I viewed the output of each push-pull on an oscilloscope. The magnitude of the each signal never exceeds 20% above the target voltage. Because sine wave AC has magnitude of over 40% above RMS, I would naively think a magnitude of 20% above RMS is perfectly acceptable. I would also naively think that a noisy power supply is fine for a filament driver.

Are these assumptions reasonable? Is this circuit an acceptable power supply for the IV-18 filament? Am I missing anything?

User avatar
phild13
 
Posts: 247
Joined: Mon Sep 10, 2012 1:05 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by phild13 »

along with the FETS, you might also try using 1N5817 (DigiKey - 1N5817FSCT-ND) for the Schottky diodes on the low voltage (5v) side of things. They have a 450mV forward drop at 1A which is much better than the original ones. You can also use (digiKey - SB160-E3/54GICT-ND) for the higher side as I think they are also better than the original ones.

I was getting about 4.98 volts unloaded when I tested the Q3 transistor board I got the other day. Unfortunately, for some reason the boards are missing all ground connections on the right hand third of the board, and so are useless. I have not put together one of your B revisions I have yet.

Have you tried looking at some of the DC/DC Boost components? They might take less space and do a better job. You could use a buck if you pull from a higher (9v) than need voltage.

User avatar
jarchie
 
Posts: 615
Joined: Sun Jun 24, 2012 2:16 pm

Re: Ice Tube Clock IV-18 To-Spec Hack

Post by jarchie »

PhilD13 wrote:along with the FETS, you might also try using 1N5817 (DigiKey - 1N5817FSCT-ND) for the Schottky diodes on the low voltage (5v) side of things. They have a 450mV forward drop at 1A which is much better than the original ones. You can also use (digiKey - SB160-E3/54GICT-ND) for the higher side as I think they are also better than the original ones.
Thanks for the tip! I'll grab some of those in my next DigiKey order.
PhilD13 wrote:I was getting about 4.98 volts unloaded when I tested the Q3 transistor board I got the other day.
When I did the to-spec hack in the original post, I suspect that I hacked the firmware to output direct current across the push-pulls, and measured the voltate unloaded when I should have measured the voltage loaded. (At the time I did not have access to a true RMS DMM or oscilloscope.) I suspect that there will be significant voltage drop when loaded... but maybe not, just my suspicion.
PhilD13 wrote:Unfortunately, for some reason the boards are missing all ground connections on the right hand third of the board, and so are useless.
What horrible luck. That sucks! :-(
PhilD13 wrote:Have you tried looking at some of the DC/DC Boost components? They might take less space and do a better job. You could use a buck if you pull from a higher (9v) than need voltage.
I haven't looked at other options, but I should be able to fit another regulator on the xmas board without a major redesign. The LM317 only needs to absorb 300 mW (3v drop at 100 mA), so I am using a T092 package with a maximum power dissipation of 625 mW. It gets a bit warm but not hot (maybe 40 degC?). Without the need to power the filament, the 7805 only needs to absorb 40 mW of power (4v drop at 10 mA). That figure rises to only ~120 mW with the Adafruit Ultimate GPS module, so I can also swap the 7805 for an equivalent TO92 part. And the FETs can be squeezed a bit closer together than the current BJTs.

I'm reasonably happy with the AC waveform from the loaded push-pulls when powered by the LM317. When active, the push-pull output voltage rises by 5v but initially overshoots by maybe 15%. My thinking is that this kind of dirty power should be fine for driving the filament... But I'd like to make sure I'm not missing something.

Are there any good reasons to provide cleaner power to the filament?

EDIT: I thought it might help if I added pictures of the voltage across the filament.
peak.gif
peak.gif (3.57 KiB) Viewed 4784 times
rise.gif
rise.gif (3.34 KiB) Viewed 4784 times
fall.gif
fall.gif (3.31 KiB) Viewed 4784 times

Locked
Please be positive and constructive with your questions and comments.

Return to “Clock Kits (discontinued)”