Now, measurements are very interesting, but a measurement says nothing if you don’t know what the measurement was done with and, perhaps more importantly, what the settings are. We walk you through our set-ups and tell you what the most important settings are.
- Prism dScope III
- CLIO 12 with QC for speaker measurements
- Wavecrest SIA3000 (jitter measurements)
- Sourcetronic 2829A LCR
- Rigol 1054Z scope
- Rigol 815 spectrum analyser with tracking generator
- Rigol 1062Z 60 MHz function generator
- Rigol DL3021 Electronic Load
- Rigol DP832 Lab power supply
- Rigol DM3058E bench multimeter
- GWinstek ASR 2050 programmable AC load
- Tekbox LISN with LISN mate for separating common mode/differential mode noise
- CON-POWER T8 CDN for network separation / measuring noise of ethernet gear
One of the reasons we have a lot of Rigol is that we prefer not to need a mishmash of software. Now we can control most of it with one kind of software – almost all our Rigol equipment has USB and/or Ethernet connectivity to enable PC-based operation – which is convenient. In addition, Rigol proves to be reliable and easy to operate.
Apart from the main equipment, a lot is needed to actually start measuring. Think of good probes (we use a 300 MHz probe from Rohde & Schwartz (10X) for the jitter measurements), dummy loads, (variable) resistors to load the power supplies and, last but not least, lots and lots of adapters and measuring cables. Here, too, quality is very important. After all: these must not form a weak link. We try to avoid inerference and therefore always keep cables as short as possible.
A few things that often recur in our measurements:
- Rohde Schwartz probes
- Tekbox Current probe
- 250 Watt dummy loads – 4 and 8 Ohm
- Mathews 400 MHz Buffer amp for converting 1 MOhm to 50 Ohm input (for the Wavecrest).
- Hirschmann cables (both measurement cables and coax / BNC).
What is nice about both Prism dScope III and Wavecrest is that they work with fixed settings. Prism works with scripts, for example. That prevents a lot of errors in measurements. After all: we are all humans. And people make mistakes.
So by using fixed scripts, measurements are comparable (within Alpha Audio, because other platforms may use different scripts or of course different equipment!), and they are also reproducible. We have seen this many times with our own equipment, which we often use as a check before measuring a review device.
Wavecrest has tried to make measuring jitter very easy with their software. Although jitter measurements are always particularly complex, because everything affects jitter: power supply, cables, adapters….
In many cases, we use our Rigol lab power supply to power a device or clock. Incidentally, we also power the buffer amp with it. That neat, little device can run on batteries, but they run empty quickly and we found that the Rigol works just as cleanly: there was no difference in result.
By using the Rigol as a fixed power supply, we have a fixed set-up for these measurements. The Rigol has proven to be very clean and stable before. An excellent basis for measurements.
However, it is important to realise that jitter measurements are very sensitive and that not every device is equally easy to measure. Sometimes a clock is not easily accessible and sometimes no good ground can be found. This affects the measurement, because the probe’s ground should be connected as close as possible to the clock. If you don’t, you will get distorted measurements pretty quickly.
We measure power supplies mainly in two areas: stability and noise. We measure noise with the LISN (Line Impedance Stabalizer), which can send noise via the BNC output to an analyser. We also have an LISN Mate that can separate common mode and differential mode. This is useful because these two forms of noise have different causes and also need to be taken care of in a different way. We discuss this in more detail in the section on power supply tests.
As an analyser, we mainly use the Prism dScope III for low-frequency noise (up to 90 KHz) and the Rigol Spectrum analyser for high-frequency (9 kHz and higher). Usually we test up to 1 MHz or if useful: up to 10 MHz. Above that is not very useful anymore.
To measure stability, we have a so-called electronic load. It can load the power supply very precisely and can also run scripts. The load is fully programmable and can put results in a file. This allows us to make nice graphs showing how the power supply handles loads. For now, we only have a load for DC power supplies.
Filters and network equipment
We are still working on a good test for filters. It is incredibly difficult to test mains filters, as they are mainly dynamic loads. That is difficult to simulate. Now, with the LISN, de new programmable AC power supply, the function generator and some ‘dirty power supplies’, we have found a way to show the effectiveness of a filter. After all, we can show the noise of that power source with and without a filter.
We can also use the Prism to show the filtering effect of a filter. And with the function generator (on white noise mode) and the Spectrum Analyzer, we can show the filtering effect in the higher range. That says nothing about the quality or musicality, but it does show whether it does anything.
We also use this set-up to measure switches in terms of noise. Especially common mode. After all, what matters most is how quiet a switch is. The LISN can show the common mode noise of the power supply. And via the CDN T8 (network decoupling), we can capture noise on the network and pass it on to the Spectrum Analyzer or pass it through the LISN mate first to separate common mode / differential mode. We discuss this in more detail in the section on networks.
Sourcetronic LCR and cables
Finally, cables. This is a very tricky subject, of course. Each cable requires a different approach. We measure a speaker cable differently from an interlink or digital coax cable. We also explain this in the chapter on cables.
In addition, we use the Sourcetronic LCR to map out the cable properties. The 2829A has the advantage that it can sweep from 20 Hz – 300 KHz via the supplementary software. This allows us to make beautiful plots and also compare cables in one graph.
Before each measurement, we calibrate the LCR so that measurements remain comparable. When we compare cables, they are of equal length. That is crucial for the measurements.
With measurements, it is particularly important that they are done under equal conditions, with equal equipment and in an equal set-up. At Alpha Audio, we have fixed scripts and set-ups for measurements. Measurements within Alpha Audio are therefore similar, unless otherwise stated.