Renovating the Thiel CS 3.5
My audio journey dates back several decades and began with the KEF 104 reference speakers followed by the Thiel 3.5 floor standers and finally resulted in my current setup with Thiel 3.7 speakers. It is hard to part with speakers that have been a wonderful companion on a great journey, but the signs of aging are undeniable. The KEF's lost their high range and probably need an upgrade of the crossovers and the Thiels midrange driver is no longer serviceable and the tweeters are beginning to show their age but are still serviceble. The Thiels come with a bass equalizer that extends the bass in the 20 range, but parts for this equalizer are harder to get and the design of the equalizer also has room for improvement. In other words, both speakers beg to be renovated and obtain a new lease on life as sitting in a closet is no way to go.
In this post I intend to document this journey for the reason of inspiring others to do the same and to get constructive feedback on ways to get better results as I imagine others have more experience in certain areas then I have. For me it is a nice integration of many different aspects (designing, measuring, constructing, listening etc.) culminating in an end product that will be a new companion for someone else and represents an upgrade but preserves the beauty of the past.
My first project will be upgrading the Thiels. To understand the essence of these speakers it is important to be be familiar with the design philosophy of Jim Thiel. He was a strong believer in the importance of time and phase coherence of the speaker and only used first order crossovers and drivers that were capable of covering a wide frequency range without noticeable distortion. He ended up custom ordering the drivers for his speakers making it hard now to find any substitues. He spend endless times modifying the crossovers to get the best listening experience with the challenging design goals. The speaker cabinets of the Thiels were designed with attention for detail and after 30 plus years are still very acceptable and have braces to keep resonances to a minimum. The cabinets were the sloped design to maintain time coherence of the three way system and require a measuring distance of 6-8 ft to allow the drivers to integrate. To be able to fine tune a new driver with the coherence design goals, I selected to use a complete digital source chain, with dsp and amplification for each driver and the options of eq for room effects. The goal was to keep the source chain at or above the 16 bits or 96 dB standard for THD and N and the most affordable way to do this was to use Fusion amps from hypex. The user advantage of streaming audio from a cell phone contributed to using a chromecast audio device as the link between the plate amps and the cell phones.The budget for the drivers replacement, amps, dsp etc was to stay under $1000 for each speaker. With all of this as a background, lets see how this project unfolds.
1. Selecting Driver replacements
For the Thiel, several midrange substitutes have been proposed and the ScanSpeak 10F/8414 is now recommended by Thiel but comparing the specs of this driver ( https://www.madisoundspeakerstore.c...canspeak-discovery-10f/8414g-10-4-full-range/) with the specs of the wavecor wf120bd04 (https://www.parts-express.com/pedoc...wavecor-wf120bd03-04-07-08-specifications.pdf) made me decide for the wavecor. The paper cone design subjectively has a similar sound quality as the original driver, and a better horizontal range near the cross over points of 300 and 2500 Hz were relevant. Also wanting to try a new brand with good reviews contributed to giving the nod to the wavecors.
The size of the wavecor is slightly larger then the original driver, allowing a mounting of the driver without any modifications. The waveguides however need some edge removal of 1/16 inch on the inside to not impede with the free movement of the driver. Mounting the driver is straight forward and connecting the wires requires basic soldering skills. I removed the crossovers and installed two more sets of bindings posts at the bottom of the speaker. The cover of the existing binding posts is easily removed and shows that the wiring for each driver from the crossover is combined at the binding posts. After removing the crossovers, the wires to the crossover from each drivers are long enough to reach the bottom of the speaker and are connected to the existing and new binding posts. To get to the crossovers, the woofer driver needs to be removed, but after removal it is all very accessible to complete the modification.
2. Measuring driver impedance and phase and some additional specs. for all drivers.
I measured the drivers in the cabinets from the binding posts to make sure everything was installed correctly with the DATS software. The mid and woofers measured fine according to specs but the right tweeter was clearly not within specs. Fortunately, I still had another tweeter that measured fine and I was able to use as a substitute.
3. Measuring frequency range of drivers.
REW V5.20 and the UMIK-1 usb mic was used to take the frequency range measures. I had experimented with several types of methods, ground measurement, near and windowed measurements, and ended up getting the best results by placing the speaker on a coffee table 3 ft of the ground in a very large space, facing the ceiling. My measurement space is about 30 by 30 ft with a sloping ceiling from 12 to 30 ft.
I took a measurement of each driver 3x times, at 1", 12" and 36". I applied a frequency dependent window to the 12" and 36" measurement with a right window of 500 ms. and width of 15 cycles. It confirmed that the right tweeter was damaged.
Tweeters, mids, and woofers FR and distortion graphs:
Distortion graphs:
Note: the measurement of mids at 1" may have been slightly off in distance of one measurement.
4. EQ-ing the drivers for linearity within crossover range.
The Thiel speaker employ first order crossover filters with a 6dB slope, and XO-point located at 300 and 2500 HZ. IN REW it is possible to EQ each speaker and bring the FR natural slopes of each speaker consistent with the XO slopes. Instead of creating a horizontal FR around the xo point of at least 1 octave and then applying the High or Low pass filter, it is possible to immediately conform the FR as if an HPF and/or LPF FR has been applied without using any XO filters. Here are the EQ graphs for highs, mids and woofs:
Ok. Now we can enter these settings in the Hypex Filter design software, that controls the DSP function of the Hypex fusion amp modules. I'm using the FA503 plate amps and the setup was straight forward using the help guide except that I had difficulties uploading the filters to the DSP because the upload function was grayed out. Turns out, the amp goes asleep after 15 minutes and was no longer visible to the computer. After loading the filters, new measurements will be made of each driver as before to see if finer adjustments were necessary.
Stay tuned
My audio journey dates back several decades and began with the KEF 104 reference speakers followed by the Thiel 3.5 floor standers and finally resulted in my current setup with Thiel 3.7 speakers. It is hard to part with speakers that have been a wonderful companion on a great journey, but the signs of aging are undeniable. The KEF's lost their high range and probably need an upgrade of the crossovers and the Thiels midrange driver is no longer serviceable and the tweeters are beginning to show their age but are still serviceble. The Thiels come with a bass equalizer that extends the bass in the 20 range, but parts for this equalizer are harder to get and the design of the equalizer also has room for improvement. In other words, both speakers beg to be renovated and obtain a new lease on life as sitting in a closet is no way to go.
In this post I intend to document this journey for the reason of inspiring others to do the same and to get constructive feedback on ways to get better results as I imagine others have more experience in certain areas then I have. For me it is a nice integration of many different aspects (designing, measuring, constructing, listening etc.) culminating in an end product that will be a new companion for someone else and represents an upgrade but preserves the beauty of the past.
My first project will be upgrading the Thiels. To understand the essence of these speakers it is important to be be familiar with the design philosophy of Jim Thiel. He was a strong believer in the importance of time and phase coherence of the speaker and only used first order crossovers and drivers that were capable of covering a wide frequency range without noticeable distortion. He ended up custom ordering the drivers for his speakers making it hard now to find any substitues. He spend endless times modifying the crossovers to get the best listening experience with the challenging design goals. The speaker cabinets of the Thiels were designed with attention for detail and after 30 plus years are still very acceptable and have braces to keep resonances to a minimum. The cabinets were the sloped design to maintain time coherence of the three way system and require a measuring distance of 6-8 ft to allow the drivers to integrate. To be able to fine tune a new driver with the coherence design goals, I selected to use a complete digital source chain, with dsp and amplification for each driver and the options of eq for room effects. The goal was to keep the source chain at or above the 16 bits or 96 dB standard for THD and N and the most affordable way to do this was to use Fusion amps from hypex. The user advantage of streaming audio from a cell phone contributed to using a chromecast audio device as the link between the plate amps and the cell phones.The budget for the drivers replacement, amps, dsp etc was to stay under $1000 for each speaker. With all of this as a background, lets see how this project unfolds.
1. Selecting Driver replacements
For the Thiel, several midrange substitutes have been proposed and the ScanSpeak 10F/8414 is now recommended by Thiel but comparing the specs of this driver ( https://www.madisoundspeakerstore.c...canspeak-discovery-10f/8414g-10-4-full-range/) with the specs of the wavecor wf120bd04 (https://www.parts-express.com/pedoc...wavecor-wf120bd03-04-07-08-specifications.pdf) made me decide for the wavecor. The paper cone design subjectively has a similar sound quality as the original driver, and a better horizontal range near the cross over points of 300 and 2500 Hz were relevant. Also wanting to try a new brand with good reviews contributed to giving the nod to the wavecors.
The size of the wavecor is slightly larger then the original driver, allowing a mounting of the driver without any modifications. The waveguides however need some edge removal of 1/16 inch on the inside to not impede with the free movement of the driver. Mounting the driver is straight forward and connecting the wires requires basic soldering skills. I removed the crossovers and installed two more sets of bindings posts at the bottom of the speaker. The cover of the existing binding posts is easily removed and shows that the wiring for each driver from the crossover is combined at the binding posts. After removing the crossovers, the wires to the crossover from each drivers are long enough to reach the bottom of the speaker and are connected to the existing and new binding posts. To get to the crossovers, the woofer driver needs to be removed, but after removal it is all very accessible to complete the modification.
2. Measuring driver impedance and phase and some additional specs. for all drivers.
I measured the drivers in the cabinets from the binding posts to make sure everything was installed correctly with the DATS software. The mid and woofers measured fine according to specs but the right tweeter was clearly not within specs. Fortunately, I still had another tweeter that measured fine and I was able to use as a substitute.
3. Measuring frequency range of drivers.
REW V5.20 and the UMIK-1 usb mic was used to take the frequency range measures. I had experimented with several types of methods, ground measurement, near and windowed measurements, and ended up getting the best results by placing the speaker on a coffee table 3 ft of the ground in a very large space, facing the ceiling. My measurement space is about 30 by 30 ft with a sloping ceiling from 12 to 30 ft.
I took a measurement of each driver 3x times, at 1", 12" and 36". I applied a frequency dependent window to the 12" and 36" measurement with a right window of 500 ms. and width of 15 cycles. It confirmed that the right tweeter was damaged.
Tweeters, mids, and woofers FR and distortion graphs:
Distortion graphs:
Note: the measurement of mids at 1" may have been slightly off in distance of one measurement.
4. EQ-ing the drivers for linearity within crossover range.
The Thiel speaker employ first order crossover filters with a 6dB slope, and XO-point located at 300 and 2500 HZ. IN REW it is possible to EQ each speaker and bring the FR natural slopes of each speaker consistent with the XO slopes. Instead of creating a horizontal FR around the xo point of at least 1 octave and then applying the High or Low pass filter, it is possible to immediately conform the FR as if an HPF and/or LPF FR has been applied without using any XO filters. Here are the EQ graphs for highs, mids and woofs:
Ok. Now we can enter these settings in the Hypex Filter design software, that controls the DSP function of the Hypex fusion amp modules. I'm using the FA503 plate amps and the setup was straight forward using the help guide except that I had difficulties uploading the filters to the DSP because the upload function was grayed out. Turns out, the amp goes asleep after 15 minutes and was no longer visible to the computer. After loading the filters, new measurements will be made of each driver as before to see if finer adjustments were necessary.
Stay tuned
