Electro-Voice T35 and T350

A Semi-detailed Analysis






Two T35/350 coils seen through a microscope.  The left is an after market replacement and the right is an original.

There are 23 turns of wire on the after market one and 25 turns on the original, assuming I counted correctly.  Another after market coil (not shown) has 21 turns.

Admittedly, EV could have made these changes over the years but when purchasing an after market replacement, one has no idea of what they will receive.  Then there's the material from which the diaphragm is made.

While these differences may seem insignificant, the pair of T-350s I have, although they sound similar under most listening conditions, pink noise will reveal the sonic difference, not to mention a frequency response curve.

While the sonic difference is subtle to un-noticeable, it becomes tantamount to a nightmare to someone like me who likes to upset apple carts.  The caveat is that 'it's dangerous to be right when the established authorities are wrong' - VOLTAIRE.



A T-350 set up for a gated frequency response.  The LMS calibrated mic can be seen in the lower left corner.  It is clamped to a device I made years ago to facilitate mounting it to a tripod.

The clamp is painted purple because it works better.  That claim is made without proof.

A brief description of the LMS is at the bottom of this page.



The mic at 30 degrees off axis and at a distance of 1 meter.

The stick on the floor next to the triangle is 1 meter long.





Again, The Usual Suspects




Smile, Guys. You're NOT on Candid Camera





None of the following curves have been smoothed.

The 3 green curves are the T-35, the black and grey are the T-350L and the red/pink ones are the T-350R

I realize this looks like a jumbled mess and one would be forever lost if it were a road map.  There are clearer curves below but it's a little less clear as to their relative  performance.

However, all vertical and horizontal scales are the same and the vertical scale had been stretched to separate the curves.  It looks a lot worse than it is.  Manufacturers sometimes compress the vertical scale to make the curve look flatter, like in the next piccy.  They do all sorts of things to make their product look better. One thing most don't do is specify the conditions under which the data was acquired.  Without the conditions, the data is essentially useless.






Both the T35 and T350 are diffraction horns which means the longer axis of the horn be vertical in order to maximize the horizontal radiation, yet the darker green curve contradicts that. It seems to have a better horizontal radiation with the horn's long axis horizontal.  This phenomenon appears in the responses of the T-350 also.

Another set of curves was run after writing this far and the difference with and without a flat baffle was negligible, about a 1.5dB variation around the band between 3khz and 9khz. but nothing that came close to the light green response.

I once asked Paul Klipsch about the 90 degree rotation of the T-35 and his reply was that it makes very little difference as high frequencies tend to beam anyway.

The above seems to verify the expression that states, "In theory, there's no difference between theory and practice.  In practice, there is."











I began to question my equipment or process.  Since I remembered running a frequency response curve on a mid-range unit in a system I designed that was plagued with response peaks above 5khz. Initially I suspected the tweeter hut much to my surprise, it was the midrange.  Even a 4th order filter couldn't handle that peak as it about an octave wide and 12dB in amplitude at the center.  A band reject filter was attempted without too much success so it was back to filter theory and the drawing board.

That curve was run by the same methods always used, gated sweep.  This curve waas run with the speaker in chamber of the three way system.  In free air, then high frequency peaks were still evident with a low frequency roll-off due to no baffle.  When I compared this curve to that of the manufacturer, I concluded there was nothing wrong here.

The manufacturer's data sheet can be found here.


The product page is here.

Parts Express






The following 3 piccies are the mic at 1/2 meter from the mouth of the horn.  The poweer delivered was 1/4 watt, which is equivalent in SPL to 1w1m.  The mic was moved closer to find out if ther horizontal radiation was affected by this proximity.  It wasn't.  Neither was it affected when the mic was moved to 2 meters.

This repetition was done to verify the unexpected results in horizontal dispersion versus horn orientation which I hope will be explained below.


1/4 watt, 1/2 meter on axis



1/4 watt, 1/2 meter 30 degrees off axis horn vertical



1/4 watt, 1/2 meter 30 degrees off axis horn horizontal



The dark green and red upper pair are 21 & 45.

The medium green & medium red are 23 & 46.

The lightest green & light pink are 25 & 47.

Ignore the crosshairs; I forgot to move it aside before capturing the screen.

The curves 45, 46 & 47 were run a day later than 21, 23 & 25.

The main oddity is the 23 & 46 pair compared to the 25 % 47 pair.  They show higher output in the octave between 5khz to 10 khz with the horn horizontal. This is contrary to theory.  However, in an attempt to explain this, the conclusion was decided that while the above may be true, the response is not very flat.  While the octave mentioned above may be only 3dB higher when the horn is horizontal, this will make the horn sound very bright due to the width of the passband, one full octave. The response is much flatter with the horn vertical.

This also shows that horizontal dispersion of this horn is essentially moot above about 12 khz.  Paul Klipsch was right when he told me in response to my asking about the horizontal orientation of the tweeter that high frequencies tend to beam and they surely do. I wonder how much, if any thought he gave to that 3dB rise in that 5khz to 10khz octave. 

Looking at the curves above 12 khz, 30 degrees off axis with the horn vertical versus horizontal shows no difference with the exception, especially with the horn vertical, of the drop of about 5 dB compared to the on axis response.  The pair of vertical orientation curves is extremely similar to that of the on axis response.  (Lightest green and light pink being close to parallel to dark green and dark red.)




This set of curves is the same as the previous with the addition of the blue ones which werwen run a day later at 1w2m, the purpose of which was to determine if distance would have any effect, other than a drop of 6dB.  It appears there was none, other than the aforementioned 6dB drop.

The upper dark blue is 48, the lower not so dark blue is 49 and the light blue is 50.  The 3dB rise in the octave between 5khz and 10khz is also evident here.  







The following are 3 data sheets for the T35 and T350.  These shown are just thumbnails to get a glimpse of the full size sheets which can be downloaded by left clicking on the thumbnail and should produce a good print.

There are two sides to each data sheet.

FYI:  To convert the no longer used EIA sensitivity rating to the current RETMA, add 49dB.

The derivation of this is shown at the bottom of the page.












Converting EIA to RETMA sensitivity rating.

The EIA is specified at 1mW at 30 feet.  (I have no idea from whose hat this was pulled.)

1mW to 1W is a 1000:1 ratio, therefore 30 dB.

30 ft.= 9.144m and 6 dB is gained or lost each time the distance is doubled or halved, therefore

taking log base 2 of 9.144 gives 3.1928.  This is how many times the distance is doubled or halved, depending on which way you go.

Multiply this by 6 (dB) as this is the change per distance doubling or halving and we get 19.16

add that to the 30dB power ratio correction and we get 49.16

Add 49 to the EIA rating of 57dB to get the RETMA rating of 106dB, 1w1m




The LMS, Loudspeaker Management System is a circuit board that plugs into an computer slot.  the unit I have was purchased in 1993 and fits into an ISA slot, no longer found.  the system is independent of the micro-processor speed as all calculations and iterations are done on the board. The only exception is the screen refresh rate which even on a 486-66 is, to all intents and purposes, instant.  The unit comes with a lab type microphone and a microphone calibration disc which only has to be installed once.  It's similar to the more modern computer based systems with the exception of it's inability to use more than one input at a time.  Therefore, it won't measure phase response directly; it's calculated.  The system is accurate to 3%.  It cost $1400 in 1993.  Linear-X has a newer version that is portable, running by a USB port.  It comes with 3 mic inputs and if memory serves me well, one or two others.  It will measure phase directly as well as make pretty waterfall curves.  The new unit sells for about $3500 with one mic.  Extra mics are about $350 each.  If I did the work now that I did 20 years ago, I'd buy it in a heartbeat but now I just can't justify the cost.



Back to the loudspeaker main page