Overhauling and Improving the Hammond M-100 Series Vibrato System

January 24, 2009

Hammond tonewheel organs beginning with the BV and CV models feature a vibrato system which modulates the frequency (and to a lesser extent, the amplitude) of the tones produced by the organ at a rate of about 6.9Hz. Unlike later all-electronic organs which modulated the tone source, the Hammond vibrato system modifies the tones themselves, on their way from the keyboard into the amplifier.

How is this possible? That was the question that Hammond engineers asked themselves in the 1930s and was answered by John Hannert’s invention of the vibrato delay line and scanner. This system works by first feeding the audio signals through the delay line, which is a series of second order passive all-pass filters. Each filter has the effect of slightly delaying the signal (about 50μs per filter stage). The length of the delay depends on where along the delay line you pick up the signal again. The delay line is the electronic equivalent of a tube into which you feed sound from one end, with a series of holes where you can listen to the sounds from. The further down the tube you listen, the more delayed the signal will be.

The second part of the Hammond vibrato system is the scanner, which consists of sixteen sets of fixed capacitor plates and one set of plates mounted on a rotor. The moving plates are interleaved between the fixed ones, much like the air-gap capacitors used in older radios. Each of the fixed plates is connected to a point on the delay line. As the rotor turns, the moving plates move past each fixed plate in turn, allowing the signal from that plate to pass into the moving plate (recall that audio signals are AC, and that AC can flow through a capacitor).

The sixteen inputs to the scanner are connected to nine of the outputs of the delay line in such a way that the rotor samples those nine outputs first from least delay to most, then back to least. In other words, if we label those output with the letters A through I, the scanning sequence is [A, B, C, D, E, F, G, H, I, H, G, F, E, D, C, B]. After one scan it starts over from A. Returning to the “long tube” analogy, this is like moving your ear along the tube, starting at the hole right beside the sound source, moving to the far end, and then back again.

In effect, scanning back and forth along the delay line is like moving toward and away from a sound source. This causes a change in frequency due to Doppler shift. Because the filters slightly attenuate the signal, there is also a change in amplitude.

Console Organ Vibrato

In a Hammond console organ (B, C, and A-100 series), there are three different vibrato settings and eighteen stages in the delay line. Using a complex multi-pole switch, the different settings connect to different sets of points along the line. At the V-1 setting, only the first nine stages are used. Twelve stages are used for V-2, skipping a few along the way. The highest setting, V-3, uses all eighteen stages, skipping half of them (more are skipped in the middle than at the ends of the delay line). All three settings have the first two stages in common.

Spinet Organ Vibrato

In the M-100 series of spinet organs such as my own M-111, there are only sixteen stages in the delay line and two vibrato settings. The small setting uses stages [0, 1, 2, 4, 6, 7, 8, 8, 8, 8, 8, 7, 6, 4, 2, 1], while the normal setting uses [0, 1, 2, 4, 6, 9, 12, 14, 16, 14, 12, 9, 6, 4, 2, 1]. The effect of these stage sequences is readily visualized by looking at a graph. The graph shows two complete vibrato cycles (corresponding to two revolutions of the scanner) at the normal and small settings. The horizontal axis represents elapsed time (in scanner steps) and the vertical axis is delay (in delay line stages):

The skipping of stages isn’t uniform, but it’s pretty close for the normal setting. However, it’s terrible for the small setting, with stage 8 used for five consecutive scanner steps in a row. This means that during those five steps, there’s no effective motion and thus no frequency shifting. It’s like the vibrato is turned off during a third of each revolution of the scanner. None of the settings on the console organs have such a plateau.

Why did Hammond do this? I don’t know. Can it be improved? Yes!

Improving the Small Vibrato

My first inclination was to come up with a small a scanning sequence that makes the (red) graph look as close as possible to a sine curve:

Notice that this one still has flat areas but they are smaller and divided equally between the top and bottom. This sequence would be ideal because it would sound like one were moving toward and away from the sound source in a cyclic manner (like sitting on a merry-go-round, listening to an organ-grinder off to one side). The problem with this sequence is that it has only one point in common with the normal setting, namely the connection to stage 0. This means an 8-pole switch would be needed to switch the remaining 8 delay line connections between normal and small, but the M-100 series has only a 4-pole switch.

I briefly considered adding some relays to get more switch positions, but since perfect symmetry isn’t required and none of the console organ vibrato settings have plateaus anyway, I came up with an alternate small sequence that can be implemented with the existing 4-pole switch. Its graph, together with the factory normal sequence, looks like this:

Notice that unlike the original small sequence, it goes to stage 10 instead of only stage 8. This places it somewhere between the V-1 and V-2 settings of a console Hammond.

Improving the Normal Vibrato

Can we make any improvements to the normal vibrato scanning sequence as well? I think so. The peak of the factory normal sequence is quite steep, resulting in a rather sudden turn-around from falling pitch to rising pitch. By slightly changing the choice of delay line points near the peak, we can smooth this out a bit, without changing any of the points we’ve so carefully kept in common with the small sequence.

Here then are the proposed new M-100 series vibrato sequences (solid thin lines) overlaid on the original factory ones (translucent wide lines):

Notice that both the normal and small curves are very close to symmetrical, and have the first five points in common. Numerically, the sequences are [0, 1, 2, 4, 6, 7, 8, 9, 10, 9, 8, 7, 6, 4, 2, 1] and [0, 1, 2, 4, 6, 10, 13, 15, 16, 15, 13, 10, 6, 4, 2, 1].

Overhauling the Delay Line

Before attempting to improve the vibrato, it is necessary to restore it to factory specifications if the line box has the older wax capacitors (as shown in the photo at the top of this article). While these were very good capacitors in their time, they don’t hold their specifications well as they age. Their capacitance actually increases over time, which has two detrimental effects in this application:

1. The cut-off frequency of the low-pass filters is reduced, meaning that a larger portion of the sound spectrum entering the delay line will be attenuated. This results in a higher percentage of amplitude modulation, giving more of a tremolo effect.

2. The delay introduced by each stage is increased, resulting in a stronger frequency modulation. To some extent, this makes up for the extra amplitude modulation (i.e. one can still perceive the vibrato within the tremolo).

Overall, these two effects combine to make the modulation too intense. I found that in the normal setting, the frequency would shift so far that it sounded like I was playing the wrong notes. There was also a very noticeable “thumping” due to excessive amplitude modulation.

The cure for this is to replace all the capacitors. In my line box, there are fifteen 0.0056μF (5.6nF) capacitors and one 0.0027μF (2.7nF) capacitor, all rated for 200V. I chose to replace these with new polypropylene capacitors of the same values rated for 630V, which I ordered from Digi-Key (part number P3506-ND for the 0.0056μF and P3502-ND for the 0.0027μF).

The first step is to remove the delay line from the organ since it is easier to work on it on a workbench. I unsoldered each of the wires connected to the delay line terminals, making note of which wire went where. Then I removed the four screws holding the delay line to the back of the upper manual. Although it’s not necessary, removing the top panel of the organ makes it easier to see what you are doing.

Once on the bench, I used a solder sucker to remove the solder where the capacitors were connected to the common bus at the bottom. I then untwisted each lead from the lower terminal. Next I heated the upper terminal and found that the upper lead would just slide out from under the inductor wires. After all the capacitors were out, I used the solder sucker again, as well as some desoldering braid, to clean excess solder off all the terminals (including the ones that the scanner wires attach to).

I measured the paired 18kΩ termination resistors and got a reading of almost exactly 9kΩ as expected. However, the 1.8kΩ resistor at the input end of the box was reading 2.2kΩ. I didn’t have a replacement 1.8kΩ 1/2W resistor handy, so I wired a 10kΩ in parallel with the existing resistor to bring the resistance back down to 1.8kΩ.

I started by installing the 0.0027μF capacitor on the far left (note that the line box is reversed relative to the schematic – the left end of the line box corresponds to the right side of the schematic). I fed one lead through the lower terminal and held the other lead against the upper terminal with an alligator clip. I then soldered the lower terminal, removed the alligator clip, and soldered the upper lead to the face of the upper terminal. I installed the remaining fifteen 0.0056μF capacitors using the same technique.

Out of curiousity, I decided to measure the capacitance of each of the old wax capacitors. The results were astonishing. Of all the 0.0056μF capacitors, the lowest capacitance I measured was 0.0079μF and the highest was 0.0107μF, a range of 41% to 91% over spec. The average was 0.0095μF, or 69% over spec. The 0.0027μF capacitor measured 0.0045μF.

After completing the recapping, I reinstalled the linebox in the organ and reconnected it exactly the way it came from the factory so I could compare the sound to how it sounded with the old capacitors. The difference was remarkable. There are audio clips at the end of this article if you want to hear for yourself.

Implementing the Improvements

Rewiring the line box for the improved vibrato delay curves is straightforward, so much so that one wonders why Hammond didn’t build it this way in the first place. The following excerpts from the M-100 schematic sum up the modification:

Notice that there are connections to non-numbered terminals on the vibrato delay line. Those terminals are already there anyway; they were just not used and not numbered in the schematic. Also note that the stage numbers I’ve been using so far in this article refer to filter stages, not terminal numbers. The easiest way to remember which filter stage is being talked about is to look at the inductor number immediately to the left of the connection in the schematic. For example, the terminal numbered 11 is stage number 9. There is no inductor to the left of terminal 1, so we’ll call that stage 0. To avoid errors, I wrote the stage numbers on the line box before I installed the new capacitors.

These are the steps I followed to carry out the rewiring:

1. Moved the wire connected to stage 14 left one position (right on the schematic) to stage 15.
2. Moved the wire connected to stage 12 left one position to stage 13.
3. Moved the wire connected to stage 9 left one position to stage 10.
4. 

   Rewired using the improved scheme.

   Unscrewed the vibrato switch block from the front panel, slid it out towards the back of the organ and turned it over. I then removed the wire between the three switch terminals that connected three steps of the scanner to stage 8 of the delay line. This left a wire connected to only one switch terminal, which turned out to be the terminal that should remain connected to stage 8.

5. Soldered two new wires to the other two switch terminals and connected them to stages 9 and 10 of the delay line (the brown and orange wires in the photo).
6. Reinstalled the vibrato switch block.

That’s all there was to it.

How Does It Sound?

In order to be able to make a comparison between the three different stages of this project (before, after recapping, and after rewiring), I made recordings each step of the way. To make it easy to do some signal analysis, each recording consists of only a single note, the A above middle C, with only the 8′ drawbar pulled out. This gives an almost pure 440Hz sine wave.

Each recording below is four to five seconds long. The graphic for each covers exactly four vibrato cycles (four revolutions of the scanner), and was taken directly from the recordings. The changes at each stage are easy to see. The difference between the old and recapped vibrato is also easy to hear, whereas the difference between the overhauled and rewired vibrato is more subtle.

I find the vibrato sounds much better than before. I can now use full vibrato while playing music and it sounds quite nice. Before the overhaul, it was most unpleasant. The recapping made the biggest improvement, but I think that the rewiring has also subtly improved the effect.

                    

Related Articles

If you've found this article useful, you may also be interested in:

• Rebuilding a Hammond AO-29 Amplifier from the Ground Up
• Hammond Organ Tonewheel Generator Capacitor Replacement and Calibration
• Overhauling the AO-29 Amplifier in the Hammond M-100 Series
• Window Seat Bookcase Tone Cabinets for a Hammond Organ
• Retronome – A Versatile Analog Drum Machine for My Hammond Organ
• Adding a Rotary Speaker to a Hammond M-111 Organ