Sansui TU-X1 done by an audio engineer
Here's the legendary Sansui TU-X1 that was modified and aligned by an audio engineer who specialize in "small signal processing". Just to clarify, I am not this audio engineer.
I received this TU-X1 for mechanical repair due to shipping damage. When I'm done fixing the mechanical issues, upgrades and alignment will be performed just like what I did to the owner's other tuners a McIntosh MR80 and Marantz 20B.
Initially, I didn't know who did the previous upgrades and alignment. A quick sleuthing at the Yahoo FMtuners group yielded the last technician or engineer in this case who apparently worked on it. It was Joseph Chow of Audio Horizons/Component Plus from Sacramento California.
One of the purpose of this page is to find out how well the performance of this TU-X1 after it has gone through an audio engineer who previously worked in the FM tuner divisions at Kenwood and Proton. I expect the workmanship to be exemplary and alignment to be top notch and bang on.
The workmanship criteria will be addressed by some detailed pictures of the internals including areas that supposedly have undergone audiophile modifications. Please see the pictures below and be the judge.
On the objective side of things, I did some measurements like stereo separation, THD (via FFT), THD+N, phase and frequency response just to see if the alignment was done properly. I've provided some comparative measurement of another tuner that used similar technology and generation for reference, the Nikko Gamma V. This particular Nikko Gamma V used as reference is a one-of-a-kind unit. Most of the knowledge I currently possess in RF and baseband was applied to this tuner. Do not expect your modded and aligned Gamma V to come close to this unit, sorry.
I work with different high end tuners from Burmester, Sansui, Kenwood, JVC, Marantz, Pioneer, Yamaha, Sony, Luxman, Aiwa, Hitachi, Nikko, Akai, Sanyo, etc. as a hobby.
Now on to our TU-X1.
Visual Inspection
TU-X1 power supply that was recapped. usual suspects like Elna, Panasonic and Nichicon caps were used. Bridge rectifier was replaced with FRED diodes. Nothing wrong with all these except with fact that the old corrosive glue used known by knowledgeable folks in the vintage audio repair world is still on the board. The corrosive glue should have been cleaned off or else the old glue will react to the new leads when it comes into contact with liquid or moisture.
closer look of the old glue that is still on the old metal film resistor body. Notice also the old glue still on the 10k carbon film resistor lead.
masking tape to hold the Teflon wires together. The extensive use of masking tape for holding and insulating wires is deplorable and in some cases a safety issue.
mix of masking tape and heat shrink tubing used to insulate the "high quality audio grade cable". This view is from the bottom side where the AM board is located. Note the dark color of the gold-plated RCA connectors. Snake oil perhaps?
FM output direct to the rear.
more masking tape, these ones are located under the MPX board. The output selector for the rear outputs were bypassed and these are the old cables left floating.
AM output with masking tape
TU-X1 MPX board. Some Nichicon MUSE and LN electrolytic capacitors. Jantzen Superior Z-caps are the red ones in tie wraps. No sign of any resistor upgrades as claimed by the seller. There are a couple of 3.9k metal film resistors beside the MPX IC HA11223W but those are original.
The $188 Audio Horizons Platinum 1A fuse! note the ding on the transformer cover that was already present before this tuner was shipped from thousands of kilometers away. It got deeper on its way to North America due to poor packing.
Visual Inspection Comments
I was out of words when I saw the internals of this TU-X1. It took me over 10 years to finally see one of Joseph Chow's work. See this link of the sales pitch of this exact TU-X1.
Objective Testing
Actual measurements of this modded and aligned Sansui TU-X1 super tuner was performed with an Audio Precision System Two 2322A and Panasonic VP-8131A. This equipment setup is state-of-the-art in terms of residual noise, residual distortion and crosstalk.
Mono THD+N
mono THD+N, spec is 0.03% @100Hz, 0.02% @1kHz and 0.04% @6kHz, As shown below 0.02% @100Hz, 0.027% @1kHz and 0.045% @6kHz. It sort of passed as the numbers are close enough. The shape is not the same though. The Gamma V shape looks more reasonable and no odd response that the TU-X1 is showing. THD+N bandwidth = <10Hz to 80kHz.
The rise in THD+N after 1kHz is due to the contribution of inherent noise in the tuner and FM system and at the same time the lower demodulated output level as the frequency goes up due to de-emphasis. FM deviation remains constant at 75kHz within limits of test equipment for all test frequencies.
Let's just assume that noise is constant across the entire frequency, THD+N includes the noise component so if the fundamental level is getting lower as one goes higher in frequency and the noise level remains constant, one would see a trend of apparent increase in THD+N as the frequency goes up even though the actual THD remains the same.
FM Stereo THD via FFT
The FM stereo tests shown were done with the left channel only, meaning the right channel has no modulating signal.
1kHz stereo left only THD is about 0.06% mainly 2nd and 3rd harmonic, spec is 0.03% failed this test. Gamma V THD is 0.02% and mainly 2nd harmonic.
100Hz stereo left only THD is about 0.1% mainly 2nd and 3rd harmonic, 3rd harmonic is much higher, spec is 0.04% failed this test. Gamma V THD is 0.03% and again mainly 2nd harmonic
The horizontal scale for the Gamma V is 1/4 less than the TU-X1 by changing the sampling frequency, a method called decimation. This was done to show a finer detail around the 100Hz fundamental. Even with the Gamma V plot showing finer detail, inter-modulation products are virtually non-existent.
5kHz stereo left only THD is about 0.03% mainly 2nd and 3rd harmonic, spec @6kHz which is close enough to 5kHz is 0.05% so it passed this one. Gamma V THD is 0.015%. If we used the conventional THD+N measurement instead of FFT, the Gamma V THD+N will be 0.1% with the 19kHz pilot leakage dominating.
The Gamma V shows much cleaner spectrum just like the previous 1kHz and 100Hz FFT plots. Very little sign of inter-modulation products. The Gamma V is a good platform, it has issues but with the proprietary re-design and careful alignment, it can be made into a fine tuner that is budget-friendly.
Despite all this stellar measurement numbers for the one-of-a-kind Gamma V, it is not immune to IBOC. I have future plans to design a passive IBOC filter (also known as post detection filter) which can be incorporated into tuners like the Gamma V, TU-9900, etc. Finally, here's the frequency response of the IBOC filter that I have designed and prototyped.
Frequency Response vs Phase
frequency response vs phase, spec is 20Hz - 15kHz +0.2dB/-0.8dB. As shown below it doesn't meet the frequency response spec at all. We have -1.4dB/-2dB. Note the excessive phase shift of +30° at 20Hz to -490° at 15kHz. No idea what causes this massive phase shift. Both low and high extremes have significant roll-off. The previous owner who flagged this TU-X1 commented that the bass is
I've never heard a tuner with as convincing low bass reproduction."
From years aligning tuners and listening to different programs off-the-air, the measured numbers contradict the claims made by the previous owner above. Thin sound is apparent for tuners that have this much low frequency attenuation. The sound is also going to be muffled due to the -2dB high frequency roll-off.
On the other hand, the reference Nikko Gamma V phase shift from 10Hz to 15kHz was 0° to 265° respectively, frequency response deviation was -0.1dB/-0.4dB.
For the astute reader, the straight slope of the low frequency roll-off of the Gamma V is due to the fact that the original frequency sweep was done in linear scale, the scale was then changed to log to make it consistent with the TU-X1 horizontal log scale.
It is rare to see a manufacturer publish phase shift response. The phase response provided here is just for reference. Phase shift is related to group delay. An electronic device that exhibit constant group delay across the frequency of interest tend to reproduce the original signal with more fidelity. Group delay is a very important parameter for data transmission and audio.
FM Stereo Separation
wide IF stereo separation, spec is 45dB @100Hz, 50dB @1kHz and 35dB @10kHz, shown below 55dB @100Hz, 49dB @1kHz and 34dB @10kHz. The TU-X1 passes this one as a few dB difference is not that critical. The top trace is for narrow IF mode. The reference Gamma V shown next has way better separation and more flat across the audio spectrum. The Gamma V stereo separation from 20Hz to 15kHz was better than 50dB and from 200Hz to 6kHz was over 60dB!
The TU-X1 and Gamma V share the same Hitachi HA11223W MPX decoder IC. Both tuners use linear phase LC filters for the IF with the TU-X1 having 3x 4-stage LC in wide mode. In narrow the TU-X1 employs an additional SAW filter in front of the 3x LC filters while the Gamma V has 3x 6-stage LC filters for a total of 18 LC stages.
The reference Gamma V has stereo separation of over 50dB from 20Hz to 15kHz in wide mode. If we compare apples to apples, the Gamma V's narrow is almost equivalent to the TU-X1 wide. The TU-X1 has 4 extra LC filtering on the front end which brings it to 16-stage LC. The reference Gamma V even in narrow IF mode, its stereo separation still beats the TU-X1 in wide IF mode. I should also add that the TU-X1 has no separate adjustment for narrow IF stereo separation, the Gamma V does like most high end tuners.
Conclusion
The alignment and modification done to this TU-X1 is not outstanding. Measurement numbers didn't meet expectations. It failed some of the parameters like frequency response and stereo THD. One can argue that this TU-X1 was out of alignment due to shipping damage which I doubt, look at the stereo separation measurements and mono THD+N measurements. If an FM stereo tuner is out of alignment, all these parameters should be off too one way or the other.
The frequency response was off. Frequency response don't change that drastically especially on the low end unless the electrolytic capacitors have dried up over the years. The sloppy response of the TU-X1 is due to blind use of boutique parts like the big film capacitors at the output and questionable choice of electrolytic capacitor replacement on the MPX and buffer stages. If this tuner was modified properly and a frequency sweep system is not available, spot check of the frequency response at 20Hz, 1kHz and 15kHz in stereo should have revealed the deficiencies. There is no excuse for not doing this.
This is what I try to tell DIY people, a collection of boutique parts, high spec op amps, etc. alone is not going to make a tuner sound good, proper alignment will yield much better results.
If you are sending your FM tuner for modification, upgrade and alignment, ask the technician or engineer about what equipment they will use, a detailed list of work to be done and a before and after measurements of key FM tuner parameters.
I have modified and aligned dozens of tuners super and not so super and from experience, the manufacturer's mono THD+N spec is easier to match than the stereo THD. Based on the above measurements, it appears that little was done to meet the TU-X1 stereo THD specifications. More emphasis was placed on installing expensive parts.
The Sansui TU-X1 is a great tuner, I bet it can be made to perform as good if not better than the reference Nikko Gamma V. The amount of work that needs to be done to restore this TU-X1 back to its original state is huge.
I have started to dismantle the rear panel.
This page was created 2017-06-20, last updated 2017-10-12