mainpicture




Astable Multivibrator Built Solely From 555s

1 Basic idea

I guess every electronics geek made their fair share of astable multivibrators with 555 ICs, and I'm no exception. When I saw advertisement for this 555 design contest, I did not pay too much attention at first. But then I got an idea: "hey, its name is 555 design contest, not 555-with-other-components-contest. Could I actually construct something using 555s as the only components in the entire design?" The result has been a resounding "yes"! And so, I'm presenting you one of my contributions to the contest - a classical astable multivibrator that is made solely from 555 ICs. No resistors, no capacitors or other components... well, at least not in their common form. And the same concept can be applied to many, many other 555-based circuits. I've chosen the astable multivibrator as the example, because everybody knows how it works.


2 Multivibrator schematic

Just for the sake of completeness, let's have a quick look at the classical 555 astable multivibrator schematic in figure 2.1.

Classical 555 multivibrator
Figure 2.1. Astable multivibrator with 555

Okay, so the question was, how can I leave out RA, RB and C from that schematic? Well, I did not leave them out, I just used other 555s to make them. Both resistors and capacitors can be actually found inside the 555 IC structure. I tried this with both bipolar and CMOS versions of the 555 IC and I suceeded both times.

Resistor replacement was pretty easy fo figure out. It is a well-known fact that every single 555 contains a voltage divider made of three resistors which is connected between pins 1 and 8. In the bipolar version, its overall resistance is 15 kΩ. But the capacitors were a big question. So for my experiments, I bought four 555 types from two different producers, 11 pieces of each type. All were in DIP packages, because breadboarding is so much easier with them compared to SMD components. In case somebody wants to repeat my experiments exactly, actual part numbers and Farnell order codes of parts I used are in table 2.1.

Table 2.1. List of 555 types used in experiments
Producer Technology Part number Farnell code
National Semiconductor CMOS LMC555CN 9488243
National Semiconductor Bipolar LM555CN 9488235
Texas Instruments CMOS TLC555CP 1103036
Texas Instruments Bipolar NE555P 9589899


3 CMOS version

In figure 3.1 you will find internal schematics of CMOS 555 which I ripped from Texas Instruments' datasheet. Here, the main voltage divider is formed by a cascade of self-biased transistors (green rectangle in figure 3.1). Its resistance is 240 kΩ in case of TLC555CP and 220 kΩ in case of LMC555CN. I measured 10 pieces of each type and differences between individual pieces were around ±2 kΩ.

Internal schematic of CMOS 555
Figure 3.1. Internal schematic of CMOS 555

But where to find a capacitor inside the 555? Well, this is why I first turned my attention to the CMOS version of the chip - theoretically, the silicon oxide layer between transistor gates and channels should behave as a quite good capacitor. Normally, this capacitance is considered a parasitic property, but is sometimes employed in useful on-chip capacitors. I have this handy RLC meter Escort ELC-131D, which can measure capacitance and loss tangent at the same time thanks to its dual display. So I measured every combination of CMOS 555 pins to see which ones have as large capacitance as possible, but at the same time, as low loss tangent as possible. My prime candidates were pins 2, 4 and 6, because they are inputs with just one transistor gate connected to them. Measurements confirmed this assumption - combinations of pins 1-2, 1-4 and 1-6 had the best results. I thought this was because pin 1 is connected with semiconductor substrate (bulk) of the chip. In case of LMC555CN, the capacitances were around 50 pF with tangent loss of 0.7. TLC555CP was even better, the capacitances were 75 pF while the tangent loss remained around 7. I was a bit surprised that I measured such high capacitances (CMOS gate capacitances are usually in the order of femtofarads), but I was thrilled that my 555-only contraption might actually work, so I did not pay much attention.

And so I built multivibrator according to figure 3.2a on a solderless breadboard - I used normal resistors as RA and RB, but I connected another CMOS 555 instead of C. And... it failed to oscillate. It worked fine with a normal capacitor, but with "capacitor 555", it did nothing. I tried everything I could think of - I varied supply voltage, I tried many different values of RA and RB, I tested every combination of "capacitor 555" pins I could devise - but the circuit always failed to oscillate.

Wrong connection of "capacitor 555"
Working connection of "capacitor 555"

Figure 3.2. Schematic of wrong connection of CMOS "capacitor 555" (a) and working circuit (b)

I was about to give up when I realized one very important detail: today, every input pin of CMOS integrated circuits is protected against ESD! This protection usually has the form of fast diodes which are connected in reverse polarity between the input pin, GND and power rail. You can see two examples in figure 3.3. I found the simpler circuit in figure 3.3a here and I copied the more complex circuit in figure 3.3b from this application note. The RLC meter takes only small-signal measurement of the capacitance, meaning that test voltage is well under 1 V. Moreover, there is no DC voltage present on the capacitor during the measurement (only expensive benchtop RLC meters can do that). Such small voltage of course couldn't engage the ESD protection diodes, so such measurements looked fine. But in 555 astable multivibrator, the voltage on the capacitor moves between 1/3 and 2/3 VCC, which is in the order of volts - which is high enough for the protection diodes to become conductive. So I had to make sure that the diodes would not engage - and they could engage only if there was voltage higher than VCC or lower than GND on any input pin. But with pin 8 floating like in figure 3.2a, the high voltage limit was not defined. In other words, I had to connect the power pin (8) of the "capacitor 555" to supply voltage, so the protection diodes would be always reverse polarized, as is illustrated in figure 3.2b. And then the circuit began to oscillate! Afterwards, I also realized why my RLC meter showed capacitances of dozens pF on every 555 input - what it actually measured were the barrier capacitances of those protection diodes. Of course, the transistor gates and wires on the chip have their capacitances too, but my guess is that they're tiny compared to diodes' capacitances. After some experiments, I found out that only pin combinations 1-2 and 1-4 can be realiably used as "capacitor 555". Pins 2 and 4 also could be connected together to make one bigger capacitance. Pin combination 1-6 worked too, but was unreliable - sometimes the circuit failed to start oscillating after powerup or there were false pulses in the output signal. My best guess is that the "capacitor 555" did not like the fast signal changes on all its inputs (2, 4, 6) at once and its internal structure behaved erratically. So if you want to try this, I recommend schematic according to figure 3.2b.

CMOS_ESD_1

CMOS ESD 2

Figure 3.3. Two examples of CMOS input pin protection circuits

After that, only one step remained - to replace normal resistors RA and RB with 555s. This went without problems and the circuit oscillated like with normal resistors... almost. There was one difference which will be explained in chapter 3.2. Note that the "resistor 555s" have to be connected so that pin 8 has positive potential in relation to pin 1.

I also found out that the circuit was somewhat unstable with +5 V power supply - large jitter (sometimes as large as 20% of the entire period) could be sometimes seen on the output signal. But this disappeared when I raised the supply voltage to +10 V.

3.1 CMOS test board

Well, the experiments on the solderless breadboard turned out pretty good, the circuit really worked in the end. But I was met with scepticism when I've shown the working circuit to my friends. I heard "nah, you have those passive components hidden inside the beadboard" or "that breadboard has so big parasitic capacitances that the capacitances inside 555s are negligible" etc. And so I decided to make a simple "test board" to prove them wrong. To make it more versatile, I designed it so up to five 555s can serve as RA, five 555s can serve as RB and ten 555s can be placed as C. Schematic of the board is in figure 3.4, its layout is in figure 3.5; its dimensions are 83x60 mm. As you can see, the only components on the board are 555s... 21 of them, to be precise.

Schematic of CMOS test board
Figure 3.4. Schematic of CMOS multivibrator test board

CMOS test board bottom layout

CMOS test board component placement
Figure 3.5. Layout of CMOS multivibrator test board

I created the board in freeware version of Cadsoft Eagle 5.11, you can download the source files here ZIP file. The board was produced by wet rapid prototyping (photomask and iron chloride etching) and drilled by hand. I soldered socket for every 555, so I could add or remove the ICs easily during my experiments. Since the "resistor 555s" are connected in series, a wire must be placed into their socket if not populated by actual 555 IC.

Photo of CMOS test board, component side

Photo of CMOS test board, copper side
Figure 3.6. Photographs of CMOS test board (click for larger images)

3.2 Measurement results

Measurements were taken with the help of Agilent DSO6032A digital oscilloscope and were done on the test board. Note that part numbers in figure 4.3 exactly match with reality - I used LMC555CN as resistors, TLC555CP as capacitors and LMC555CN as the actual multivibrator IC (I tried other combinatons of part numbers as well, but I got similar results). Using the oscilloscope, I measured times of "high" and "low" states on the multivibrator output (IC1 pin 3) and also the overall frequency. These results are in table 3.1. Measurements were done with +10 V supply voltage.

Table 3.1. CMOS multivibrator frequency for different number of "passive component 555s"
555s serving as RA 555s serving as RB 555s serving as C thigh
[μs]
tlow
 [μs]
freq
[kHz]
1 1 1 11.4 0.15 87.0
1 1 2 12.4 0.15 80.6
1 1 3 13.8 0.15 72.5
1 1 4 15.2 0.15 65.8
1 1 5 16.8 0.15 59.3
1 1 6 18.3 0.15 54.5
1 1 7 19.8 0.15 50.4
1 1 8 21.3 0.15 46.9
1 1 9 22.7 0.15 44.0
1 1 10 24.2 0.15 41.2
2 1 10 43.0 0.30 23.3
2 2 10 54.8 0.60 18.3
3 2 10 69.8 0.60 14.3
3 3 10 82.4 0.60 12.1
4 3 10 104 1.50 9.62
4 4 10 108 1.50 9.30
5 4 10 128 1.50 7.84
5 5 10 127 1.50 7.91

Here is video where I demonstrate that the multivibrator really works. Note that measured frequencies in the video are a little different from those in table 3.1; this is because I made those measurements in different days.



Oscilloscope screenshot of multivibrator waveforms is in figure 3.7. Yellow trace is output signal (IC1 pin 3) and green trace is voltage on the "capacitor 555s". The figure shows these signals with one 555 as RA, one 555 as RB and one 555 as C (configuration according to the first row in table 3.1). Note that measured  frequency in table 3.1 is different from the frequency in figure 3.7; this large difference was caused by parasitic capacitance of the oscilloscope probe (Agilent 10073C) when I connected it to the "capacitor 555". Therefore, all measurements in table 3.1 were taken without the probe on the capacitor.

Oscilloscope screenshot with 1 RA, 1 RB and 1 C
Figure 3.7. Output and capacitor waveforms of multivibrator with with 1 RA, 1 RB and 1 C

As you can see from the results, the thigh time and frequency nicely scales with the number of "capacitor 555s." On the contrary, the tlow time is always very short (the output signal has over 99% duty cycle) and stays short even when more RB "resistor 555s" are added. This happens due to presence of the internal ESD protection diodes. When "capacitor 555" discharges during tlow time, reverse current flows through RB and voltage on it also reverses. But since RB is actually another 555, its ESD diodes open... and of course, an open diode has much lower resistance than 220 kΩ. I was even able to measure diode's drop voltage with a diode tester (it was around 0.6 V as expected). So in reality, the multivibrator behaves more like equivalent circuit in figure 3.8.

AFF equivalent circuit
Figure 3.8. Equivalent  circuit representing actual behavior of the multivibrator

4 Bipolar version

After the success with CMOS version, I turned my attention to bipolar 555s. Figure 4.1 shows their internal schematics as can be found in National Semiconductor's datasheet. Again, the main voltage divider is easy to see. Now the question was if there were some components that could be used as capacitors.

Internal schematic of bipolar 555
Figure 4.1. Internal schematic of bipolar 555

Bipolar version has no ESD protection diodes, so their barrier capacitance could not be used. Therefore, I tried to find some other reverse-polarized PN junctions, so I could employ them as the capacitor C. When pin 1 is connected to ground, base-collector junctions of transistors Q7 and Q14 are reverse polarized, which meant that pins 2 and 7 could be used for that purpose. So I made test circuit according to figure 4.2a on breadboard and amazingly, the circuit oscillated! Resistors RA and RB have value of 15 kΩ in this circuit, because I planned to replace them with another bipolar 555s. After a few experiments, I found out that pin 4 can also be used as a capacitor - while it wasn't immediately apparent from the internal schematic, its base-collector junction is too reverse polarized. So if you want to try this, you can use schematic in figure 4.2b, it has the highest capacitance I could find inside bipolar 555. 

First try of bipolar "capacitor 555"

Final version of bipolar "capacitor 555"

Figure 4.2. First testing schematic (a) and final version (b) of bipolar "capacitor 555"

Unfortunately, even the combined capacitance of all three pins is very small, it was under 10 pF in all 555s I measured. This, coupled with much lower RA and RB values (compared to CMOS), resulted in high output frequencies in the range of hundreds of kHz. Unlike the CMOS version, the bipolar multivibrator worked reliably at +5 V supply voltage and was still operating even when I decreased it to about 3.5 V.

4.1 Bipolar test board

The bipolar test board is very similar to the CMOS one, practically the only difference is in connection of "capacitor 555s". The board is slightly smaller (83x56 mm) than its CMOS counterpart, because there are less copper traces on it.

Schematic of bipolar test board
Figure 4.3. Schematic of bipolar test board

Bipolar test board bottom layout

Bipolar test board component placement
Figure 4.4. Layout of bipolar test board

You can download Eagle source files of this board here ZIP file. Some people asked me why there are short pieces of wire (Nets) connected to pin 8 of every "capacitor 555". Well, Eagle has this "smart" function that it automatically connects all power pins which have the same name. To prevent this behavior, I had to connect every pin 8 to a different signal, which was easiest done by drawing a short net to it.

Photo of bipolar test board, component side

Photo of bipolar test board, copper side
Figure 4.5. Photographs of bipolar test board (click for bigger images)

4.2 Measurement results

I made the same measurements as with the CMOS version, the only difference was that I used +5 V supply voltage this time. Measured results are in table 4.1. As you can see, the internal capacitance of bipolar 555s is really low - every added "capacitor 555" caused only slight decrease of the output freqency. Adding "resistor 555s" had more significant impact, but compared to the CMOS version, the output frequencies were still 5 to 10 times higher.

Table 4.1. Bipolar multivibrator frequency for different number of "passive component 555s"
555s serving
as RA
555s serving
as RB
555s serving
as C
thigh
[μs]
tlow
 [μs]
freq
[kHz]
1 1 1 0.98 0.13 905
1 1 2 1.05 0.13 855
1 1 3 1.10 0.11 826
1 1 4 1.17 0.12 775
1 1 5 1.23 0.12 740
1 1 6 1.31 0.12 699
1 1 7 1.38 0.13 667
1 1 8 1.44 0.13 639
1 1 9 1.49 0.12 623
1 1 10 1.56 0.11 601
2 1 10 4.53 0.11 216
2 2 10 7.12 0.8 126
3 2 10 9.85 0.8 93.9
3 3 10 11.6 2.0 73.5
4 3 10 13.8 2.0 63.3
4 4 10 16.1 5.0 47.4
5 4 10 18.7 5.0 42.2
5 5 10 20.8 7.6 35.2

Oscilloscope screenshots of output waveforms are in figure 4.6. Again, yellow traces are output signals and green traces are voltage on the "capacitor 555s". Figure 4.6a shows output signals with one 555 as RA, one 555 as RB and one 555 as C. Figure 4.6b shows output signals with five 555s in series as RA, five 555s in series as RB and ten parallel 555s as C (in other words, all sockets on the test board were populated). This represents the first and the last row in table 4.1, respectively. Measurements of the bipolar version were too affected by oscilloscope probe parasitics, but their influence was not as grave as in the CMOS version. 

Oscilloscope screenshot with 1 RA, 1 RB and 1 C

Oscilloscope screenshot with 5 RA, 5 RB and 10 C

Figure 4.6. Output waveforms with 1 RA, 1 RB, 1 C (a) and with 5 RA, 5 RB, 10 C (b)

Times tlow are again very short so the bipolar version behaves more like circuit in figure 3.8, too. But every added RB prolongs it aproximately by 1.5 μs, so the output signal duty cycle moves between 70 and 90% (the CMOS version always had over 99%).


Conclusion

Well, the good news is that both versions of the multivibrators really work! However, I guess their usefulness is limited - especially the CMOS version produces signal with duty cycle over 99% and the circuit is somewhat sensitive to touch or other movements of foreign objects in its proximitiy. The CMOS version is capable of producing low freqencies around 8 kHz - quite a surprise, considering that one "capacitor 555" value is on the order of pF. The bipolar version can produce only much higher frequencies, but its output waveforms look somewhat better.

Now I have to devise a meaningful use for those 44 pieces of 555 ICs I bought for the experiments...