PCI-6602 - What to do with unused counter inputs?

Question

I've got an application where I need to track the position on two quadrature encoders.  I've developed my application with VB6 however, I've been using a sample LabView program called "Meas-Ang-Pos-Buff.vi" for testing purposes.  (I got the example off the NI discussion board).  I'm seeing some very odd results that I can't explain. 
For testing purposes, I've got one encoder tied to multiple channels on my 6602.  I have the same 'A' signal and the same 'B' signal (from the one encoder) jumpered to multiple encoder channels.  I'm using a onboard pulse generator to generate the latch pulse (connected to the Index/Z of each counter) to ensure there is no skew in my count values.  I'm using an external servo to turn the encoder at a rate equal to my application speed.
The unexplanable results I'm seeing are below:
If I'm monitoring counters 0, 1 and 2, I see the following.
1)  If I have counters 0,1,2 and 3 inputs connected to the same 'A' & 'B' from my encoder, the counts on all 4 channels (0,1,2,3) track perfectly.
2)  If I disconnect counter 3 and only have counters 0,1, and 2 connected to the encoder, then counter 0 and 2 track perfectly, but counter 1 looses pulses (only when traveling in one direction, the other direction seems to be fine)
3)  If I connect counter 3 to ground and have counters 0,1, and 2 connected to the encoder, I get the same result - bad.
4)  If I connect EVERY counter input (2,3,4,5,6) to ground and only monitor counters 0 and 1, I still get bad results.
In summary, I ONLY am able to see the same number of counts while monitoring 3 channels IF 7 of the 8 counters are connected together (the 8th being my latch).
This happened to me in the past on a previous setup that did the same thing, but I was able to fix it by moving from counter 1 to counter 4.  At that time, I chalked it up to a bad counter 1, but I think there is a bigger issue since it is happening again.
Please any advice would be appreciated.

Answer 1

Thanks for the response.  This one has really got me confused and any help is greatly appreciated.
To answer your questions, no I have not modified the sample program and I'm using NI DAQ v8.5.
I've decided to break the problem down to see if I could find a pattern, but no such luck.  Here is the testing that I've done.
I'm making the following connections while using the sample program I mentioned earlier:
Counter 0...............PFI_39 (term 02) - Channel_A(0).................PFI_37 (term 40) - Channel_B(0)
Counter 1...............PFI_35 (term 07) - Channel_A(1).................PFI_33 (term 06) - Channel_B(1)
Counter 2...............PFI_31 (term 34) - Channel_A(2).................PFI_29 (term 66) - Channel_B(2)
Counter 3...............PFI_27 (term 31) - Channel_A(3).................PFI_25 (term 63) - Channel_B(3)
Note:  I connect the "Z" channel for gating purposes in my VB application, but this isn't required in the LabView sample program since the gating appears to be done internally.
My testing results:
1.  If I jumper my one encoder to channels 0,1,2, & 3 at the same time, everything works great.  Each channel reports the same number of counts regardless of whether I'm turning the encoder in the positive or negative direction.
2.  If I disconnect channel 3 and only have 0,1 & 2 connected, then 0 & 2 track perfectly, but channel 1 looses counts in the negative direction.
3.  If I only connect channels 0 & 1, then channel 0 counts perfectly in both the pos and neg direction, but channel 1 looses counts in the negative direction.
4.  If I connect any channel by itself (0,1,2 or 3), each one counts perfectly.
5.  If I connect channels 0 & 4, everything seems to be fine.
Like I said in my previous e-mail, if this hadn't already happened to me in a previous application, then I would chalk it up to a bad board.  If I didn't get good results on each individual channel then I would chalk it up to a bad encoder pulse train.  However, it definitely seems to have something to do with connecting multiple counters at the same time in a Motion Encoder Application.  I have read the 6602 User Manual backwards and forwards looking for a nugget of explanation, but have found none.  There was a section entitled "Counter Input Selections" on page 4-12 (table 4-2) that might offer some explanation, but neither me or my colleagues could understand what was being described.  Any help deciphering that would be appreciated.
As of now, I'm going to proceed with my application, using channels 0 &4 (just as I've done in the past), however I don't feel comfortable not having an explanation for the 6602's behavior.  Any insight would be greatly appreciated.