Mearuring crosstalk for adjusting azimuth.
In follow-up to a thread which started last week about setting azimuth to optimise crosstalk, I would like to present my opinions on how this should be done.
Crosstalk is a measure of the ability of a cartridge to resolve the left and right channels separately. A groove which has modulations in only one channel should produce a signal in the same channel of the cartridge outputs. However, due to imperfections in cartridge design, this is not the case: there is always some leaked signal in the other channel – this is called crosstalk. Ideally the leaked signal should be a very small percentage of the main signal. Poor crosstalk (when the leaked signal is significant relative to the main signal) will result in poor stereo separation, as musical instruments which are recorded in one channel will be simultaneously reproduced in the other channel (albeit at a lower level).
If one imagines the motion of the stylus, when viewing the cartridge from the front, the cause of cross talk can be easily visualised. A cartridge comprises of a pair of coils, one for the left channel, and one for the right channel. In the ideal case, the generating coils of a cartridge are positioned so that motion of the stylus along a line at an angle which is 45 degrees clockwise relative to the vertical would give rise to a change in the magnetic flux threading the left channel coil of the cartridge, but no change in the magnetic flux threading the coil connected to the right channel. Hence this motion would produce a voltage at the left channel output only; there would be no voltage across the right channel output. Similarly, motion of the stylus along a line at an angle which is 45 degrees anti-clockwise relative to the vertical, should produce a change in the magnetic flux threading the coil for the right channel, but no change in the flux threading the coil connected to the left channel. So this motion would produce a voltage at the right channel output only.
In practice, the above case of zero crosstalk is impossible to achieve, so even when the motion of the stylus occurs along either one of those 45 degree lines, there is still some signal leaked into the wrong channel. Leakage may be due to non ideal alignment of the coils, or due to the fact that it is impossible to design a cartridge so that motion of the stylus along a line at 45 degrees to the vertical produces a change in the flux threading only one of the coils.
It has been argued that the azimuth alignment of the cartridge should be adjusted to minimise crosstalk.
To do this, the normal procedure requires a test record with separate left and right 1kHz tones. The azimuth is firstly adjusted so that the stylus is perpendicular to the record when viewed from in front. The crosstalk is measured in both channels, and the azimuth is then re-adjusted until the relative level is the same for each channel. For example supposing that after the first test, the leaked signal was 1% into the left channel, and 0.5% into the right channel, in that case, the azimuth would be re-adjusted so that the leaked signal was 0.75% of the main signal in both channels.
The argument in favour of this method of setting azimuth, is that it optimises stereo separation. However, it should be pointed out that this method may result in a sub optimum alignment of the stylus in the groove, which may cause increased record wear. Further arguments against this method are based around the fact that the azimuth varies with VTA for tonearms with non aligned bearings (such as SME3009) and for uni-pivot tonarrms, so that setting azimuth for minimum crosstalk with a test record, results in optimum azimuth only for discs of the same thickness as the test record. Further problems occur if the platter profile is dished, or if the vertical axis of the toneram pivot is not exactly perpendicular to the record surface.
Nonetheless, despite arguments to the contrary, many people still believe that azimuth should be adjusted for minimum crosstalk - three dimensional imaging is often quoted as one of the strong points of vinyl reproduction compared to CD reproduction.
So to the main point of my post here. How should the crosstalk be measured for the propose of adjusting azimuth?
According to the Stereophile May issue (I can’t remember if it was April or May) the crosstalk is calculated by playing back one of the single channel tones (say the 1kHz left channel tone) and measuring the voltage at the left speaker terminals and then measuring the leaked voltage at the right speaker terminals (ideally zero) and calculating the relative level in dB.
The formula for calculating relative level in dB is given below.
20*log(V2/V1) where V1 is the main or reference voltage and V2 is the leaked voltage.
The procedure is repeated for playback of the track with the recorded signal in the other channel, and the crosstalk for that channel is calculated using the same formula..
The azimuth is then re-adjusted until equal crosstalk results for both channels.
I disagree with this method of calculating crosstalk. It does not take into account the effect of channel imbalance on the crosstalk measurement. In effect this method assumes that a cartridge has perfect channel balance.
Scanning through a list of cartridge specs shows that channel imbalance is typically specified to within 1 to 2dB (see list below)
Now consider the case of a cartridge with a 1.5 dB channel imbalance (such as the Ortofon Rohmann) and which has perfect coil alignment but where there is a leaked signal of 2% into each channel. The leakage does not give rise to the same voltages in each channel due to the imbalance of 1.5dB. Using the above method of calculating crosstalk the result would be –35.5dB and –32.5dB. A difference of 3dB between the channels. The better number -35.5dB would correspond to the louder channel. Consequently a correction would be made to the azimuth despite the fact that the alignment is optimum.
Where the user has a balance control on the pre-amp, channel imbalance can be rectified very easily by adjusting the balance until the directly measured signal in each channel is the same. But doing this changes the crosstalk measurements above, and would require a recalculation, and a yet another re-adjustement of azimuth.
Clearly it would be better to calculate the crosstalk in such a way that channel imbalance is ignored. In fact, it is very easy to do this and the procedure is almost the same as that described above.
As before, an LP track with a 1kHz signal in one channel (say the left) is played. The signal produced in the left channel for this track is measured and the voltage is recorded as V1. Then the track with modulations on the right channel is played back, and the new signal in the LEFT channel is measured and recorded as V2.
The crosstalk is calculated using the same formula as before: 20*log(V2/V1)
This procedure is then repeated with the measurements performed on the speaker terminals of the right channel. Right channel modulated track - V1, left modulated track - V2.
The two crosstalk measurements are compared, and if there is some difference, the azimuth is adjusted.
To summarize the difference between the two methods: in the first case, a single track was used, and the two voltages ware measured from different channels, whereas in the second case two tracks were used, and the voltages were measured from a signle channel.
The benefit of the second method is that channel imbalance is automatically cancelled from the equations, so in the case of the perfectly aligned Ortofon Rohmann above, the corsstalk would measure as –34dB in both cases, and no adjustment would be made.
Of course, the second method proposed above will not correct the channel imbalance, however azimuth has almost no effect on channel balance, and compensating for channel imbalance by adjusting azimuth will certainly result an a poor alignment. As already mentioned, for those using pre-amplifiers with a balance adjustment, channel imbalance can easily be corrected. For those without a balance control, then the imbalance will remain, but can be corrected by adjusting the listening position.
I hope this explanation is of some benefit.
Some specs on channel imbalance for well known cartridges.
MC Jubilee: 1dB
MC Rohmann: 1.5dB
DV XX-2: 1dB
DV10x4 mk I: 1dB
DV 20X: 1dB
Reading back over the above post, I must confess to having given the wrong impression in the procedure.
Whichever method is used for measuring crosstalk, the aim is not necessarily to have an equal result for both channels. Equal crosstalk in both channels would be optimum of course, however not if achieving that was at the expense of the absolute result in either channel.
The aim of adjusting azimuth for minimum crosstalk is to achieve the lowest figure for crosstalk in whichever channel gives the worst result.
For example –29db left and –35dB right is a better result than–26dB left, and –26.5dB right, despite the fact that the difference between the left and the right is greater in the first case.
Sorry about the seemingly contradictory recommendations. I still recommend the second method for measuring the crosstalk.
Thanks for the info Bkearns. I used the Cardas record for the 1kHz left and right tones. My measurements into a Fluke 77 volt/ohm meter are:
Measurements were made with the speakers connected and amp volume at about 1/4 turn. Is it as simple as trying to balance these numbers by means of azimuth adjustment. I read your post several times and I'm a bit confused about the means of arriving at the dB. For example:
20 x (46 / 583) = 1.578. Is this correct? Sorry if I seem like an idiot. I'm trying to get it right.
The voltages which you posted would correspond to a crosstalk of -22dB for the first pair and -23dB for the second pair.
Your mistake was not taking the log of the V2/V1 before multiplying by 20.
-22 to -23dB is an acceptable result, though a very good figure would be closer to -30dB Whether your figurs are optimum depends the cartridge you are using.
You could try adjusting the azimuth one way or the other to see if a larger difference in the voltages results.
But if you are happy with the set-up as it is, then please remember the following adage: “If it ain’t broke, don’t fix it”!
Nice system. Think I'll try it.
We're of course assuming a perfect amplifier here by measuring V at the speaker terminals.
I tried that, but the results were not so good. See the following post.
The problem is that the leaked signal at the phono preamp output should ideally be only a few mV, so you would need a screened cable to connect from the phono output to the meter, otherwise the voltages will be hidden in RFI.
I agree that it is not a perfect method.
You are definitely the Master when it comes to these kinds of analyses. I agree wholeheartedly with your method. I only wish I had thought of it, myself.
Unfortunately, I’m too lazy to adjust azimuth electronically so I simply adjust the cartridge to be parallel with the record surface. How much improvement will I notice by doing the azimuth job properly?
This post is made possible by the generous support of people like you and our sponsors: